JEL Class: K14, K41
TOWARD AN ECONOMIC THEORY OF PRO-DEFENDANT CRIMINAL PROCEDURE
By: Keith N. Hylton† & Vikramaditya S. Khanna††
Abstract
In this paper we provide a justification for the pro-defendant bias in American
Criminal Procedure that we argue paints a more complete picture of the extent and
breadth of these pro-defendant procedures than the most commonly forwarded
justifications to date. The most commonly forwarded rationale for the pro-defendant
bias in American Criminal Procedure is that the costs associated with false convictions
(i.e., sanctioning and deterrence costs associated with erroneous imposition of non-
monetary or criminal sanctions) are greater than the costs associated with false
acquittals. We argue that on closer inspection this rationale does not, by itself, justify the
extent of our pro-defendant criminal procedures. We offer another justification for these
protections – to constrain the costs associated with abuses of prosecutorial or
governmental authority. In a nutshell, our claim is that these procedural protections
make it more costly for self-interested actors, whether individuals or government
enforcement agents, to use the criminal process to obtain their own ends. The theory
developed here explains several key institutional features of American Criminal
Procedure and provides a positive theory of the case law as well.
† Professor of Law, Boston University School of Law; J.D., Harvard Law School; Ph.D. (Economics),
M.I.T.
†† Associate Professor of Law, Boston University School of Law; Visiting Associate Professor of
Law, Harvard Law School, Spring 2001; S.J.D Harvard Law School.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 1
TOWARD AN ECONOMIC THEORY OF PRO-DEFENDANT
CRIMINAL PROCEDURE
By: Keith N. Hylton† & Vikramaditya S. Khanna† †
© 2001, Keith N. Hylton & Vikramaditya S. Khanna. All rights reserved.
I. INTRODUCTION
For many people who study the American criminal justice system the
procedural protections offered to defendants seem likely to increase the crime
rate.1 At the simplest level, these protections permit some factually guilty
defendants to escape conviction, which should increase the incentives of
wrongdoers to engage in criminal acts.2 This seems odd given that criminal
wrongs are considered the most serious wrongs in society and hence the ones we
should most wish to reduce.3 What might then explain the seeming willingness
† Professor of Law, Boston University School of Law; J.D., Harvard Law School; Ph.D. (Economics),
M.I.T.
† † Associate Professor of Law, Boston University School of Law; Visiting Associate Professor of
Law, Harvard Law School, Spring 2001; S.J.D. Harvard Law School. The authors would like to thank John
Coates, Dhammika Dharmapala, Jesse Fried, Louis Kaplow, Reinier Kraakman, Michael Meurer, Mitch
Polinsky, Mark Roe, Steven Shavell, and William Stuntz for their helpful discussions and suggestions. We
would also like to thank Michelle Carlucci, Adam Forchheimer, Jessica Fritz, Angie Nguyen, Nicholas
Oldham, Seema Srinivasan, Paula Uscilla, and Andrew Yang for excellent research assistance. In addition,
many thanks to the John M. Olin Center for Law, Economics, & Business at Harvard Law School for funding
support while Professor Khanna was visiting at Harvard Law School in Spring 2001.
1 See, e.g., Raymond A. Atkins & Paul H. Rubin, Effects Of Criminal Procedure On Crime Rates:
Mapping Out The Consequences Of the Exclusionary Rule, (Oct. 23, 1998) (unpublished manuscript available on
file with authors) (finding Miranda may have increased total crime rates by eleven percent and violent
crimes rates by thirty-three percent); Paul G. Cassell, The Guilty & The “Innocent”: An Examination Of Alleged
Cases Of Wrongful Conviction From False Confessions, 22 HARV. J.L. & PUB. POL’Y 523 n.30 (1999)[hereinafter
Guilty & Innocent]; Paul G. Cassell, Miranda’s Social Costs: An Empirical Reassessment, 90 NW. U.L. REV. 387,
451 (1996)[hereinafter Social Costs] (finding after Miranda criminal suspects are less willing to confess to their
crimes). But see, John J. Donohue, III, Did Miranda Diminish Police Effectiveness, 50 STAN. L. REV. 1147 (1998)
(dissecting the statistical analyses in Cassell’s work and criticizing the use of statistics in measuring the
import of Court decisions).
2 See Miranda’s Social Costs, supra note 1, at 485.
3 See S.E. Marshall & R.A. Duff, Criminalization and Sharing Wrongs, 11 CAN. J.L. & JURIS. 7, 7 (1998)
(the term “criminal” indicates a serious condemnation of an activity or action); Susan Estrick, Rape, 95 YALE
L.J. 1087, 1183 (1986) (“conduct is labeled ‘criminal’ in order to announce to society that these actions are not
to be done and to secure that fewer of them are done”); Henry M. Hart, Jr., The Aims of The Criminal Law, 23
LAW & CONTEMP. PROBS. 401, 404-05 (“[w]hat distinguishes a criminal from a civil sanction . . . is the
judgement of community condemnation which accompanies and justifies its imposition”). See also John. C.
Coffee, Jr., Does “Unlawful” Mean “Criminal”?: Reflections on the Disappearing Tort/Crime Distinction in
American Law, 71 B.U.L. REV 193, 194 (1991) (“the factor that most distinguishes the criminal law is its
operation as a system of moral education and socialization . . . [a]s a result, the criminal law often and
necessarily displays a deliberate disdain for the utility of the criminalized conduct to the defendant”);
Toward an Economic Theory of Pro-Defendant Criminal Procedure 2
of American law to permit wrongdoers to escape conviction and thereby
potentially increase the crime rate?
Many commentators have offered a variety of reasons to support “pro-
defendant” procedural protections.4 In particular, many have argued that in the
criminal process we should be more concerned about the social costs generated
by a false conviction (e.g., diluted deterrence and the costs associated with
putting someone in prison wrongfully) than the similar costs generated by a false
acquittal.5 This rationale, which we term the “traditional error-cost rationale”,
suggests that because the costs of false convictions are greater than false
acquittals we should be willing to reduce false convictions even if that leads to an
increase in false acquittals and a net increase in the total number of errors. 6 In In
Re Winship, the Supreme Court adopted this rationale as the primary justification
for the high standard of proof in criminal trials.7 We, however, argue that for the
traditional error-cost rationale to justify our current criminal procedure system
we would need to make Herculean assumptions about the frequency and costs of
false convictions relative to false acquittals, and the evidence is simply not there
to support such assumptions.8 Consequently, the traditional error-cost rationale
is unlikely to provide a good justification for the extent and magnitude of the
pro-defendant bias in American criminal procedure.
In this paper we offer another justification for these procedural
protections: to constrain the costs associated with abuses of prosecutorial or
governmental authority. We argue that this justification gives a more complete
picture of why it might be desirable for criminal procedure to be strongly biased
Sanford H. Kadish, Excusing Crime, in BLAME AND PUNISHMENT: ESSAYS IN THE CRIMINAL LAW 87 (1987)
(“Criminal conviction charges a moral fault . . .”; HERBERT PACKER, THE LIMITS OF THE CRIMINAL SANCTION
(1968).
4 See JOSHUA DRESSLER, UNDERSTANDING CRIMINAL PROCEDURE, (2d ed. 1997) (discussing various
pro-defendant biases and their justifications); Donald J. Boudreaux & Adam C. Pritchard, Civil Forfeiture and
the War on Drugs: Lessons from Economics and History, 33 SAN DIEGO L. REV. 79 (1996); Randolph N. Jonakait,
Biased Evidence Rules: A Framework for Judicial Analysis and Reform, 1992 UTAH L. REV. 67, 68 (arguing that pro-
defendant evidence rules are defensible as they promote justice).
5 See RICHARD A. POSNER, ECONOMIC ANALYSIS OF LAW 604 – 05 (5th ed., 1998).
6 See In Re Winship, 397 U.S. 358, 372 (1970) (Harlan, J., concurring) (noting that the reasonable
doubt standard is used in criminal trials because society views false convictions as being far worse than false
acquittals); Donald A. Dripps, People v. Simpson: Perspectives On The Implications For The Criminal Justice
System: Relevant But Prejudicial Exculpatory Evidence: Rationality Versus Jury Trial And The Right To Put On A
Defense, 69 S. CAL. L. REV. 1389, 1418 (1996) (“[the reasonable doubt standard] strikes the balance very much
in favor of [increasing] false acquittal[s]” as opposed to false convictions).
7 See In Re Winship, supra note 6, at 363 – 64 (Brennan, J.)(stating that “[t]he accused during a
criminal prosecution has at stake interests of immense importance, both because of the possibility that he
may lose his liberty upon conviction and because of the certainty that he would be stigmatized by the
conviction. Accordingly, a society that values the good name and freedom of every individual should not
condemn a man for commission of a crime when there is a reasonable doubt about his guilt”), 372 – 74
(Harlan, J., concurring) (noting that the reasonable doubt standard is used in criminal trials because society
views false convictions as being far worse than false acquittals largely due to liberty and reputation costs).
8 See infra Part III.C.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 3
in favor of the defendant. Thus, this paper offers (depending on one’s point of
view) an alternative rationale for, or a significant modification of, the traditional
error cost argument in favor of a strong pro-defendant bias in American
Criminal Procedure.
In a nutshell, our claim is that criminal procedural protections make it
more costly for self-interested actors, whether individuals or government
enforcement agents, to use the criminal process to obtain their desired ends.
Absent some constraint, prosecutors and government agents might be tempted
to use the criminal process to benefit themselves or their constituents. History
provides us with a number of examples of this.9 Procedural protections impose
constraints and make the criminal process more costly to use, thereby providing
enforcement agents (e.g., prosecutors), and those who would lobby them, with a
disincentive to use the criminal process for selfish ends.10 This saves resources
that otherwise would be eaten up in the lobbying process.11 In addition,
constraining this sort of behavior is likely to enhance deterrence. The reason is
that when it is easy to enforce the law in selective ways, enforcement agents will
come under pressure to sacrifice deterrence objectives for distributive goals (i.e.
enforcing the law in ways favorable to a particular group).12 The effects on
deterrence and the direct costs associated with lobbying provide a more
complete rationale for the existence and extent of our criminal procedural
protections and a better positive theory of the criminal procedure case law. In
particular, the theory we develop here provides a better explanation for the
existence and specific form of key rules and institutional features: the reasonable-
doubt and double-jeopardy rules, restrictions on excessive and retroactive
punishments, features of the right to a jury, such as the unanimity requirement
and peremptory challenges, and others. Our theory is also corroborated by
empirical evidence on corruption from several countries.
Part II begins by providing a brief description of some core criminal
procedural protections that currently inhabit our jurisprudence. These include
the beyond reasonable doubt standard of proof, double jeopardy protections,
9 See infra Part IV.
10 As the criminal process becomes more expensive to use then prosecutors and those who lobby
them would prefer to substitute less costly methods of obtaining their desired ends (e.g., lobbying
legislatures for particular kinds of laws that disproportionately burden another group). Whether this
substitution is desirable and how we might contain rent-seeking in these other spheres is outside the scope
of this paper. We do, nonetheless, briefly discuss this issue at infra text accompanying notes 192 – 195. For a
recent discussion of the role of politics and the criminal law, see William J. Stuntz, The Deep Politics of
Criminal Law, Draft 2001 (on file with authors).
11 See, e.g., GORDON TULLOCK, THE ECONOMICS OF SPECIAL PRIVILEGE AND RENT SEEKING 96, (Gordon
Tullock ed., Kluwer Academic Publishers 1989); CHARLES K. ROWLEY ET. AL., THE POLITICAL ECONOMY OF
RENT-SEEKING 465-478, (Charles K. Rowley et al. eds., Kluwer Academic Publishers 1988).
12 In addition, an increase in perceived selective enforcement is likely to reduce the stigmatic effect
of the law and increase enforcement costs. See infra Part V.B.2.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 4
and others.13 A critical point to note at this juncture is that all of these
protections have a cumulative effect of biasing the criminal process in favor of
the defendant.14 This raises the fundamental question of whether such a bias can
be justified.
Part III inquires into the traditional rationales provided for these core
criminal procedural protections. Because the core protections in our view are
those that impart a pro-defendant bias to the law, we focus on the reasonable
doubt standard as the quintessential procedural protection. We argue that the
traditional error-cost rationale does not appear, by itself, to justify the reasonable
doubt rule. We conclude that other rationales should be examined in order to
provide a stronger basis for the current scope of pro-defendant protections.
Part IV argues that one other rationale for these procedural protections is
that they constrain the behavior of enforcement agents and make it more difficult
for such agents to use the criminal process to facilitate wealth extraction or to
benefit themselves.15 If prosecutors were allowed to easily use the criminal
process to obtain their own selfish ends then individuals and groups would
lobby the prosecutor to bring (or not bring) certain kinds of cases.16 Legal history
is riddled with examples of such behavior, which generates significant costs for
society.
Part V discusses some of these costs. Lobbying efforts are often wasteful
from society’s perspective and generate additional costs, such as a dilution in the
deterrent effect of criminal prohibitions.17 These costs are significant and
warrant consideration of methods to contain them.
13 Our reason for choosing these protections is that they seem to have the most historical support
and most directly influence the probability of being punished or the actual punishment meted out. Other
procedural protections (e.g., the Fourth Amendment’s Unreasonable Search & Seizure) do not carry the
same kind of historical pedigree and also do not impact the probability of being punished as directly as
those listed in the text. See Benton v. Maryland, 395 U.S. 784, 794 (1969) (noting that “the double jeopardy
prohibition of the 5th Amendment represents a fundamental ideal in our constitutional heritage”); In re
Winship, supra note 6, at 361 (1970) (noting that “the reasonable doubt standard is accepted as the measure
of persuasion by which the prosecution must convince the trier of all the essential elements of guilt”).
14 See W. William Hodes, Reform: Lord Brougham, The Dream Team, And Jury Nullification of the Third
Kind, 67 U. COLO. L. REV. 1075, 1078 nn.7 & 46 (1996) (noting that a criminal trial favors the defense because
of the many procedural protections); William J. Stuntz, The Uneasy Relationship Between Criminal Procedure
and Criminal Justice, 107 YALE L.J. 1 (1997) (noting that these constitutional protects favor the defendant, but
that there is still much discretionary power left with a prosecutor).
15 This is essentially an argument that the procedural protections work as a method of
reducing/constraining agency costs much like incentive schemes in the corporate context. See infra Part V.
16 See Jeffrey Standen, Plea Bargaining in the Shadow of the Guidelines, 81 CAL. L. REV. 1471, 1473
(1993) (noting that “prosecutors have an incentive to discriminate against particular defendants or
subgroups of defendants by attempting to settle like cases differently depending on defendants’ personal
characteristics unrelated to culpability”).
17 See TULLOCK, supra note 11, at 5, 96.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 5
Part VI examines some methods of constraining self-interested
prosecutors. These include procedural protections, restrictions on penalties, and
other ways to limit prosecutorial abuse of the criminal process.18 We conclude
that the procedural protections appear to work, in ways consistent with our
approach, as constraints on enforcement agents that are not fully replicated by
other methods.
Part VII applies the analyses from Parts III through VI to examine some
aspects of criminal procedure and constitutional law. In particular, the
reasonable doubt standard of proof, certain aspects of double jeopardy
protection, the right to a jury trial, the ex post facto rule, entrapment, and a host
of others.19 Part VII also provides empirical evidence based on corruption data
from several countries that is consistent with our analysis. Part VIII concludes
with suggestions for future research and analysis.
II. SOME CORE PRO-DEFENDANT CRIMINAL PROCEDURAL PROTECTIONS
There is a vast panoply of procedural protections attached to the criminal
process in the U.S.20 To focus our analysis we hone in on the handful of
protections that appear to impose a significant pro-defendant bias in the criminal
law process.21 In particular, we focus on the reasonable-doubt standard of proof,
double jeopardy protections, and the right to a jury trial. We reserve for later
18 See David Friedman, Why Not Hang Them All: The Virtues Of Inefficient Punishment, 107 J. POL. &
ECON. 259, 262 - 63 (1999) (inefficient punishments); Ronald A. Cass & Keith N. Hylton, Antitrust Intent, 74
S.CAL. L. REV. 2001 (forthcoming).
19 See Connally v. General Construction Co., 269 U.S. 385, 391 (1925) (stating that “a statute which
either forbids or requires the doing of an act in terms so vague that men of common intelligence must
necessarily guess at its meaning and differ as to its application, violates the first essential of due process of
the law”); United States v. Russell, 411 U.S. 423 (1973) (court reaffirmed support for a subjective version of
the federal entrapment defense).
20 See generally, Lewis v. United States, 518 U.S. 322 (1996) (discussing the right to trial by jury, but
deciding against permitting a jury trial on the facts of that case); Scott v. Illinois, 440 U.S. 367, 370 (1979)
(right to counsel); Miranda v. Arizona, 384 U.S. 486 (1966) (secures suspects’ privilege against self-
incrimination).
21 In our view, rules that directly impose a pro-defendant bias are those that reduce either the
probability of conviction or the severity of the punishment. See Standen, supra note 16, at 1519 (“[Better
rules] would limit opportunities for manipulation by prosecutors in their charging, and would require
criminal offenders who wish to lower or eliminate their expected punishment to alter their behavior either
to conform to the law or cause less harm.”). Rules that merely restrict the type of evidence that can be
presented, such as the exclusionary rule, do not fall within our definition of core pro-defendant protections.
This is because their impact on the probability of conviction is not as direct as the reasonable doubt standard
and also because they do not have the same kind of historical pedigree that the reasonable doubt standard
has. See Lawrence H. Tribe, Constitutional Calculus: Equal Justice Or Economic Efficiency?, 98 HARV. L. REV. 592,
607 (1985) (“[E]xclusionary rule cases…are today treated as occasion for the assessment of the marginal
deterrent effects of excluding particular categories of evidence”). Also note that it might be easier for police
and other government agencies to satisfy some parts of the Fourth Amendment (e.g., giving a Miranda
warning) as compared to satisfying the reasonable doubt standard. See Charles Weisselberg, Saving Miranda,
84 CORNELL L. REV. 109 (1998) (advocating a return to the “original” vision of Miranda).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 6
discussion, in Part VII, other protections that also impact the criminal process in
favor of the defendant.
The reasonable-doubt standard requires that the moving party (i.e., the
prosecution) prove that the defendant is guilty, beyond a reasonable doubt, of
the criminal offense(s) with which he is charged.22 Although the reasonable-
doubt formulation seems to have first appeared in 1798,23 the notion that the
standard of proof in criminal trials should favor defendants appears to have
ancient origins. Blackstone, in his description of the criminal process, noted that
“all presumptive evidence of felony should be admitted cautiously: for the law
holds, that it is better that ten guilty persons escape, than one innocent suffer.”24
Coke, considerably earlier, said that “the evidence against a prisoner should be
so manifest, as it could not be contradicted.”25 In 1970, the Supreme Court
endorsed this by holding in In Re Winship that the due process clause protects the
defendant against conviction except upon proof beyond a reasonable doubt.26
The reasonable-doubt standard stands in contrast to the “preponderance
of the evidence” standard,27 used most frequently in non-criminal cases and for
sentencing issues in criminal proceedings.28 It requires that the moving party
prove that the defendant is liable on the preponderance of the evidence or, put
simply, is more likely liable than not. 29 The preponderance rule is considered
the easier standard for the moving party to meet relative to the reasonable doubt
standard.30 Sometimes the preponderance rule is assumed to require that the
decision-maker be 51% certain that the defendant is liable before finding against
him, whereas the reasonable doubt standard is assumed to require that the
decision-maker be somewhere in the range of 90% to 95% certain before
22 See In Re Winship, supra note 6, at 361.
23 See C. MCCORMICK, EVIDENCE § 341, at 576 – 78 (1992).
24 4 WILLIAM BLACKSTONE, COMMENTARIES, at 358.
25 Id, at 349 – 50.
26 See In Re Winship, supra note 6, at 364.
27 See Ethyl, infra note 31, at 28 n. 58 (noting that different levels of certainty are required to meet the
distinct burdens).
28 See Concrete Pipe and Products Of California, Inc. v. Construction Laborer’s Pension Trust for
Southern California, 508 U.S. 602, 622 (1993) (noting that preponderance of the evidence is the “most
common standard in the civil law”); Apprendi v. New Jersey, 530 U.S. 466 (2000) (noting that the
preponderance of the evidence standard can be used for sentencing as long as the sentence is not more
severe than the statutory maximum for the offense established by the jury’s verdict); U.S. v. Lombard, 72
F.3d 170, 176 (1st Cir. 1995) (noting that the preponderance of the evidence can be used in sentencing issues
[enhancements in this case]).
29 See U.S. v. Mandanici, 205 F.3d 519, 532 (2d Cir. 2000) (Kearse, J. concurring) (“a preponderance
means more likely than not”).
30 See Martin v. U.S., 277 F.2d 785, 786 (5th Cir. 1960) (noting that the preponderance of the
evidence standard is a lesser burden than proof beyond a reasonable doubt).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 7
convicting the defendant.31 This imparts a significant pro-defendant bias in the
criminal law.
Another procedural protection is Double Jeopardy.32 The prohibition
against Double Jeopardy stems from the 5th Amendment to the U.S. Constitution,
which states “nor shall any person be subject for the same offense to be twice put
in jeopardy of life or limb”. 33 In many respects this protection is similar to the
doctrines of Res Judicata and Collateral Estoppel that are found in non-criminal
cases. 34 However, there are some differences. In particular, one difference that
has garnered much attention is the rule that normally prohibits prosecutorial
appeals of initial trial acquittals, but permits defense appeals of initial trial
convictions.35 This asymmetry in appeal rights appears like it has a pro-
defendant bias whereas in the non-criminal side the analogous doctrines (e.g.,
Collateral Estoppel) do not present such asymmetry in appeal rights.36
31 See Ethyl Corp. v. EPA, 541 F.2d 1, 28 n. 58 (D.C. Cir. 1976) (“It may be that the ‘beyond a
reasonable doubt’ standard of criminal law demands 95% certainty [internal cites omitted]. But …, a
preponderance of the evidence demands only 51% certainty.”); Brown v. Bowen, 847 F.2d 342, 345-46 (7th
Cir. 1988) (suggesting that certainty of 90% or more is sufficient to meet the reasonable doubt standard).
There is yet another standard of proof that might be used in some instances. The clear and convincing
evidence standard is used in some non-criminal contexts often as a method of determining if greater-than-
compensatory damages should be awarded in a particular case. See id., at 346. This standard occupies an
intermediate position between the preponderance and the reasonable doubt standards. See, e.g., Anderson v.
Liberty Lobby, Inc., 477 U.S. 242, 271 (1986); Cornell v. Nix, 119 F.3d (8th Cir. 1997). Although these three
standards are sufficient for purposes of our analysis it is worth noting that in the past there have been other
standards. In fact, there was at one time a standard even higher than the reasonable doubt standard. See
Anthony M. Morano, A Reexamination of the Development of the Reasonable Doubt Rule, 55 B.U.L. REV. 507, 511-
19 (1975) (arguing that prior to the articulation of the reasonable doubt standard, a higher standard existed
that required proof beyond any doubt). Although we have discussed three standards of proof, it is possible
that other standards could exist as well. There is, in theory, a continuum of standards of proof, but in
practice only three. See V.S. Khanna, Corporate Criminal Liability, 109 HARV. L. REV. 1477, 1516 n. 210; See also
Lee, infra note 66, at 25; See also MCCORMICK, supra note 23, at 378. See also, Sullivan v. LTV Aerospace and
Defense Co., 82 F.3d 1251, 1260 (2d Cir. 1996) (“The familiar burden of proof standards occur along a
continuum with ‘preponderance of the evidence’ at one end and ‘beyond a reasonable doubt’ at the other.
The ‘clear and convincing evidence’ standard falls somewhere in between.”).
32 See United States v. DiFrancesco, 449 U.S. 117 (1980); George C. Thomas III, An Elegant Theory of
Double Jeopardy, U. ILL. L. REV. 827 (1998); Akhil Reed Amar & Jonathan L. Marcus, Double Jeopardy Law After
Rodney King, 95 COLUM. L. REV. 1 (1995).
33 U.S. CONST. amend.V.
34 See Robin W. Sardegna, No Longer In Jeopardy: The Impact Of Hudson v. U.S. On The Constitutional
Validity Of Civil Monetary Penalties For Violations Of The Securities Laws Under The Double Jeopardy Clause, 33
VAL. U. L. REV. 115, 117 (1998) (noting that the civil procedure doctrines of res judicata and collateral
estoppel are similar to prohibitions of the double jeopardy clause).
35 See Vikramaditya S. Khanna, How Does Double Jeopardy Help Defendants?, Discussion Paper No.
315, Discussion Paper Series, John M. Olin Center for Law, Economics, & Business, Harvard Law School
(2001); U.S. v. Ball, 163 U.S. 662, 671 (1896); Joshua Steinglass, The Justice System in Jeopardy: The Prohibition on
Government Appeals, 31 IND. L. REV. 353 (1998); Kate Stith, The Rise of Legal Error in Criminal Cases: Some
Consequences of the Asymmetry in the Right to Appeal, 57 U. CHI. L. REV. 1 (1990); Peter Westen & Richard
Drubel, Toward a General Theory of Double Jeopardy, 1978 SUP. CT. REV. 81.
36 See Cooper v. Federal Reserve Bank of Richmond, 467 U.S. 867, 874 (1984) (noting that if the
requirements are met both res judicata and collateral estoppel are available to plaintiffs and defendants);
Saulpaugh v. Monroe Community Hosp., 4 F.3d 134, 144 (2nd Cir. 1993) (noting that “we recognize that
arguments based on collateral estoppel or res judicata may be available to either side”).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 8
Other protections that are relevant in the criminal context are the right to a
jury trial,37 the ex post facto punishment rule,38 and the excessive punishments
prohibition.39 Yet other protections exist that appear to have a pro-defendant
bias (some of which we address later in Part VII), but the crucial point for now is
that the criminal process appears biased in favor of the defense. This raises the
basic question – what justifies this sort of bias in the criminal process?
III. TRADITIONAL JUSTIFICATION FOR THE CORE CRIMINAL PROCEDURAL
PROTECTIONS: THE REASONABLE-DOUBT STANDARD
The reasonable-doubt standard presents the quintessential case of a pro-
defendant protection. The most common justification given for it is that in the
criminal process we are more concerned with false convictions than false
acquittals and hence should prefer a pro-defendant bias (as that reduces
convictions, which might also reduce false convictions).40 We will refer to this
below as the “traditional error-cost” rationale for pro-defendant protections. We
conclude that although this rationale carries some persuasive force it is not
strong enough, by itself, to justify the reasonable-doubt standard.
To show this, we approach the issue in a number of steps. We begin by
trying to determine what kind of cost differential between false convictions and
false acquittals we would need to justify the reasonable doubt standard over the
preponderance standard using traditional error-cost arguments. We then
describe and examine the factors that comprise the social costs of false
convictions and false acquittals. Traditional error-cost arguments focus on the
deterrence costs and sanctioning costs of both errors. After this we try to
determine whether the empirical evidence on the cost differential (in terms of
deterrence and sanctioning costs) between false convictions and false acquittals
seems to justify society’s choice of the reasonable doubt standard over the
preponderance standard. We conclude that the answer to this question is most
likely no, and hence we examine other justifications for pro-defendant
protections in the remainder of the paper.
A. Traditional Error-Cost Analysis
Blackstone’s claim that “it is better that ten guilty persons escape, than one
innocent suffer”41 captures the essence of the traditional error-cost justification,
37 See U.S. CONST. AMEND. VI.
38 See U.S. CONST. ART. I, § 9, CL. 3.
39 See U.S. CONST. AMEND. VIII.
40 See 4 WILLIAM BLACKSTONE, COMMENTARIES *358; see also In Re Winship, supra note 6, at 372
(Harlan, J., concurring).
41 See 4 WILLIAM BLACKSTONE, COMMENTARIES *358.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 9
which is that false convictions generate greater social concern than false
acquittals.42 In light of this a basic question is raised – how much of a difference
in the costs of false convictions and false acquittals is needed to justify the
reasonable doubt standard? Would a minor difference suffice (e.g., a false
conviction costs society $10 and a false acquittal costs society $9) or would
something more significant be needed (e.g., a false conviction costs $100 and a
false acquittal costs $5) before the reasonable doubt standard was justified?
1. Basic Framework
Perhaps the first to examine this question was John Kaplan in his classic
article “Decision Theory and the Factfinding Process.”43 Defining the standard of
proof as the probability of guilt necessary for a rational jury to convict, Kaplan
showed that the proof standard depends only on the ratio of the cost of a false
acquittal to the cost of a false conviction. To be precise, if we let P represent the
proof standard and R the ratio of the false acquittal cost to the false conviction
cost (or ratio of error costs), Kaplan showed that
P > 1/(1+ R) .44
Kaplan’s formula implies that the preponderance rule is appropriate
whenever the cost of acquitting a guilty person is equal to the cost of convicting
an innocent person (R = 1).45 The reasonable-doubt rule can also be justified in
terms of Kaplan’s formula, since the formula implies that the proof threshold
should be increased as the cost of a false conviction rises relative to that of a false
acquittal.46
42 See In Re Winship, supra note 6, at 372; POSNER, supra note 5, at 605 (noting that the social cost of
false convictions outweigh the costs of false acquittals).
43 See John Kaplan, Decision Theory And The Factfinding Process, 20 STAN. L. REV. 1065 (1968).
44 See id., at 1071 – 72.
45 See id., at 1072. Kaplan’s formula for determining the probability necessary to convict tells us one
important lesson from decision theory. The probability-of-guilt threshold that minimizes the overall costs of
error depends only on the ratio of error costs. The absolute cost of a false conviction or false acquittal is not
important in this analysis. The fact that false convictions may seem more costly for one type of criminal
defendant than another is not an important datum under the decision-theoretic approach. The key question
is whether the cost of a false conviction relative to the cost of a false acquittal is higher for one type of
defendant than for another.
46 If the ratio of error costs (R) is 1/3, so that each false conviction is three times as costly as a false
acquittal, the probability of guilt necessary to convict becomes .75. If a false conviction is four times as costly
as a false acquittal, the probability of guilt necessary to convict becomes .8. Blackstone’s expression of the
reasonable-doubt standard, that ten guilty men should be allowed to go free in order to avoid convicting
one innocent, (see 4 WILLIAM BLACKSTONE, COMMENTARIES *358) implies an error cost ratio less than 1/10 and
means that that the probability threshold must exceed 90%. Assuming 95% is a good approximation, this
requires a cost ratio of 1/19. Thus, in order to accept Blackstone’s description of the proof threshold as
consistent with that of a rational jury, we must believe that the cost of a false conviction is nineteen times
that of a false acquittal. To be precise, with a 10 to 1 cost ratio we obtain: 1/[1 + 1/10] or 1/[11/10] or
simply 10/11 (i.e., .909090… or approximately 91%). As this seems like an odd percentage we will opt for a
percentage that covers this cost ratio. 95% does, but 90% does not. Another reason for choosing 95% is that
Toward an Economic Theory of Pro-Defendant Criminal Procedure 10
Though enormously valuable as a clear statement of the assumptions
underlying alternative proof standards, Kaplan’s approach does not tell us
whether we should prefer the reasonable-doubt to the preponderance standard.
Kaplan’s approach tells us which percentage proof requirement is to be preferred
over all others given a particular ratio of error costs. However, we are choosing
not from a continuum of proof requirements, but from only two – either the
preponderance or the reasonable-doubt threshold. To determine which of these
two standards is preferable, we will need to take a different approach.
2. When to prefer the reasonable doubt over the preponderance standard
To choose between the reasonable-doubt and preponderance standards,
we will take advantage of an approach suggested by Gordon Tullock. The
relationship between error probabilities and standards of proof can be illustrated
with the diagram in Figure 1.47 The vertical axis measures the probability of
guilt. The horizontal axis measures the amount of evidence (of guilt). The
straight line captures the functional relationship between the probability of guilt
and the quantity of evidence against the defendant. The vertical line PE reflects
the preponderance-of-evidence standard. If the amount of evidence is below (to
the left) of the PE line, the defendant will be found innocent, and if the evidence
is above the PE standard then the defendant will be found guilty. Error
probabilities under the PE standard are shown by the areas OBE and PBA in
Figure 1. The probability of a false acquittal is given by OBE, to the left of the PE
standard. The probability of a false conviction is given by PBA, to the right of
the PE standard. Figure 1 also shows the same relationship under the
reasonable-doubt standard, where the vertical line labeled RD reflects the
reasonable-doubt standard.
it agrees with the traditional statistical standard for hypothesis testing, which requires the rejection of a
hypothesis when the evidence against could have occurred by chance in only one out of twenty trials. The
standard approach would suggest that we should convict the defendant – i.e., accept the hypothesis that he
is guilty – when the evidence suggests there is no more than a five percent chance that the evidence against
him is all coincidental. See THAD W. MIRER, ECONOMIC STATISTICS AND ECONOMETRICS (3d ed. 1995).
47 See GORDON TULLOCK, THE LOGIC OF THE LAW 65 (1987 ed.). Tullock uses a logistic or ‘S’ shaped
curve instead of a linear one that we use. We chose the linear shape because of expositional ease and
because using the linear shape, over the logistic one, should make it easier to justify the reasonable doubt
standard. Proof available upon request from authors. If even on the easier linear standard we cannot justify
the reasonable doubt standard then it is unlikely to be justified on the ‘S’ shaped logistic standard.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 11
A C
P
D
B
PROBABILITY
OF GUILT
E F
O PE RD
EVIDENCE
Figure 1
Two rather intuitive points follow from the diagram: the probability of a
false acquittal and the overall likelihood of error are both larger under the
reasonable-doubt standard. In moving from the preponderance to the
reasonable-doubt standard, we reduce the probability of a false conviction by the
area ABCD, and we increase the likelihood of a false acquittal by the larger area
BDEF.48 It follows that if the social costs of false acquittals and false convictions
are equal society should prefer the preponderance standard to the reasonable-
doubt rule.
If the costs of false convictions and false acquittals are not equal,49
examining the overall probability of error will be insufficient to tell us whether
the reasonable-doubt rule is preferable to the preponderance rule. In general,
society should prefer the reasonable-doubt standard to the preponderance
standard only if the expected social costs are lower under the reasonable-doubt
standard. Thus, the reasonable-doubt standard is preferable if the change in the
probability of a false conviction (ABCD in Figure 1) multiplied by the cost of a
false conviction exceeds the incremental false acquittal probability (BDEF in
Figure 1) multiplied by the cost of a false acquittal. Put simply, the reasonable-
doubt rule is preferable if the reduction in false conviction costs exceeds the
increase in false acquittal costs. Should one believe this holds?
48 See id.
49 See POSNER, supra note 5, at 605.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 12
3. Toward an Empirical Test
To determine whether, as an empirical matter, a move from the
preponderance to the reasonable-doubt standard would reduce false conviction
costs by a larger amount than it increases false acquittal costs, we must formulate
in precise terms what we mean by the costs of false convictions and false
acquittals. In the standard economic analysis of enforcement, the costs of false
convictions and false acquittals can be categorized as either “deterrence costs” or
“sanctioning costs”.50 Specifically, false convictions and false acquittals are
socially costly to the extent that they lead to sub-optimal deterrence – that is, to a
deterrence level either above or below the social optimum.51 Sanctioning costs
are the losses society bears by punishing a convicted offender, such as the social
costs of maintaining prisons, the forgone labor of the convicted offender, and the
deprivation of his liberty.52
Because a false acquittal involves no punishment at all, the only costs we
need concern ourselves with are deterrence costs. In the case of false convictions,
we need to consider both deterrence and sanctioning costs. Thus, the costs we
are concerned with are the incremental deterrence costs of false convictions, the
incremental deterrence costs of false acquittals, and the incremental sanctioning
costs. To be precise, let Dfc = the incremental deterrence cost of one false
conviction, let Dfa = the incremental deterrence cost of one false acquittal, and let
Cc = the incremental social costs of imposing sanctions.
We are now in a position to state a rough empirical test to determine
whether the reasonable-doubt rule is preferable to the preponderance standard.
The reasonable-doubt rule is preferable only if the incremental sanctioning costs
under the preponderance rule exceed the incremental deterrence costs (or
“underdeterrence” costs) under the reasonable-doubt rule. In technical terms,
and using Figure 1, the reasonable-doubt rule is preferable only if (∆)Cc >
(BDEF)Dfa – (ABCD)Dfc, where ∆ is the incremental probability of conviction. 53
50 See POSNER, supra note 5, at 604 – 05; Thomas R. Lee, Pleading And Proof: The Economics Of Legal
Burdens, 1997 B.Y.U.L. REV. 1, 5 (1997) (noting that “error costs are the social costs associated with erroneous
legal judgments. Erroneous legal judgments include decisions for undeserving defendants (Type I errors)
and decisions for undeserving plaintiffs (Type II) errors.”); Peter Wendel, A Law and Economics Analysis of the
Right to Face-to-Face Confrontation Post-Maryland v. Craig: Distinguishing the Forest from the Trees, 22 HOFSTRA
L. REV. 405, 412-13 (1993) (stating that the risk of erroneous legal judgement, or cost of error, is the
probability of an erroneous conviction times the cost of erroneous conviction, plus the probability of an
erroneous acquittal times the cost of an erroneous acquittal).
51 See POSNER, supra note 5, at 604 – 05; Lee, supra note 50, at 26 n.79.
52 See POSNER, supra note 5, at 604 – 05; Gary S. Becker, Crime and Punishment: An Economic Approach,
76 J. POL. ECON. 169, 180 (1968); Wendel, supra note 50, at 426.
53 Note that if P is the share of guilty individuals in the population, the incremental conviction
probability, ∆, is equal to P(BDEF) – (1-P)(ABCD).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 13
The empirical evidence does not directly address this question, but it does
suggest that this necessary condition is unlikely to obtain.54 The most direct
source for such evidence is provided in Anderson (1999).55 If we use, as a
conservative definition of the aggregate harm from crime, the sum of losses due
to injuries and property theft (including fraud), Anderson’s study suggests the
annual aggregate harm due to crime is roughly $1,177 billion.56 If we measure
sanctioning costs by adding the opportunity costs of the inmate’s time while
locked up plus the costs of maintaining inmates in prison, Anderson’s study
suggests that the annual sanctioning cost for all convictions is $71 billion. 57 From
these numbers one can see that the aggregate costs of crime, which can be treated
as “underdeterrence costs,” are on the order of 15 times greater than the
sanctioning costs associated with all convictions. In view of the sheer magnitude
of this differential, it seems improbable that the savings from a measure that
Also note that Cc is simply the sum of the sanctioning costs associated with the incremental false
convictions (Cfc) and incremental correct convictions (Ccc) generated by switching to the preponderance rule.
One might expect that Cfc per case is greater than Ccc per case.
Since BDEF > ABCD, the interesting theoretical question generated by this condition is whether Dfa
> Dfc – i.e., whether the deterrence costs of false acquittals exceeds the deterrence costs of false convictions.
If so, then the incremental deterrence costs of the reasonable-doubt rule are definitely positive, and it is an
open (empirical) question whether the sanctioning costs under the preponderance rule exceed these
incremental deterrence costs. As a theoretical matter, it is straightforward in this framework to show that
Dfa should be larger than Dfc. First, both are positive because increases in either type of error reduce the
difference in payoffs for complying and not complying with the law, see, e.g., Louis Kaplow & Steve Shavell,
Accuracy in the Determination of Liability, 37 J.L. ECON. 1, 5 (1994). Second, Dfa > Dfc, as a theoretical matter for
the following fundamental reason: in a well functioning enforcement system, enforcement efforts will be
targeted at guilty actors. Given this, changes in false acquittal probabilities will far outweigh changes in
false conviction probabilities in terms of their influence on incentives. For an independent formal analysis
that reaches the same conclusion, see Henrik Lando, The Optimal Standard of Proof in Criminal Law When Both
Fairness and Deterrence Matter, SSRN Working Paper available at
http://papers.ssrn.com/paper.taf?abstract_id=238334. Both the framework here and Lando’s are easily
reconcilable with Craswell and Calfee’s important article on the deterrence effects of error. See Richard
Craswell & John E. Calfee, Deterrence and Uncertain Legal Standards, 2 J. L. ECON. & ORG. 279 (1986). For an
analysis of the deterrence effects of error in the negligence context, see Keith N. Hylton, Costly Litigation and
Legal Error Under Negligence, 6 J. L. ECON & ORG. 433 (1990).
54 See David Anderson, The Aggregate Burden of Crime, 42 J. LAW & ECON 611 (1999); John J. Donohue
III & Peter Siegelman, Allocating Resources Among Prisons and Social Programs in the Battle Against Crime, 27
J.LEGAL STUD. 1 (1998). See also UNITED STATES DEPARTMENT OF JUSTICE, BUREAU OF JUSTICE STATISTICS, Costs of
Crimes to Victims (1994) (setting out the costs of specific crimes to victims, such as crimes of violence, rape,
robbery, assault, theft, and motor vehicle theft).
55 See Anderson, supra note 54. The sanctioning costs of the death penalty are not discussed here
because of the number of executions each year. See Hashem Dezhbakhsh, Paul H. Rubin, & Joanna Mehlhop
Shepherd, Does Capital Punishment Have a Deterrent Effect? New Evidence from Post-moratorium Panel Data,
Draft 2001 (on file with authors), at 29 (providing a table of the number of executions over the last 23 years
which appear to be around 40 to 50 a year). For discussion of the value of life in different contexts see W.
Kip Viscusi, The Value of Life in Legal Contexts: Survey and Critique, 2 AMERICAN LAW & ECONOMICS REVIEW 195
(2000).
56 See Anderson, supra note 54, at 629 (providing this number which represents the sum of ‘Risks to
life and health’ and ‘Transfers’).
57 See id., at 620, 624 (providing this figure which represents the sum of “Crime-Induced
Production: Corrections” and “Criminal lost workdays: in prison”).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 14
reduces the costs of crime, such as moving to the preponderance standard,
would be swamped by a rise in sanctioning costs.
For example, if shifting from the reasonable-doubt to the preponderance
standard improved deterrence by 30% it would reduce the annual cost of crime
by about $352 billion.58 To disfavor such a move, we would need to believe that
the sanctioning costs would rise by more than $352 billion – i.e., from roughly
$71 billion to $423 billion. This would be equivalent to a six-fold increase in the
prison population.59 If the reduction in crime costs were 20 % (saving about $235
billion), we would need sanctioning costs to increase from $71 billion to $306
billion to offset the gain (a little over a four-fold increase in the prison
population). If the crime reduction benefit were only 10 % (saving about $117
billion), we would need an increase of sanctioning costs to $188 billion to offset
the gain. While it is impossible to rule out these scenarios, they each require the
combination of a relatively modest impact on deterrence and an extremely large
increase in sanctioning costs, on the level associated with at least a tripling of the
prison population.60 However, the empirical evidence on the responsiveness of
crime rates to changes in prison population suggests that this combination is
unlikely to be observed.
The empirical evidence on the elasticity of crime with respect to
incarceration (i.e., the rate at which crime drops as incarceration/prison
population increases) puts the figure in a range from roughly .15 to .30.61 In
other words, a one percent increase in the prison population results in a
58 We do not expect all crime to be avoided by such a change because the standard of proof may not
be the only factor influencing crime rates and the optimal crime rate may not be 0. See Jonathan M. Karpoff
& John R. Lott, Jr., The Reputational Penalty Firms Bear from Committing Criminal Fraud, 36 J.L. & ECON. 757
(1993); Anderson, supra note 54, at 616 (finding benefits of crime due to the fact that crime induces economic
production); Eric Rasmusen, Stigma and Self-Fulfilling Expectations of Criminality, 39 J. L. & ECON. 519, 533
(1996). For factors besides the standard of proof that might influence crime rates see also Peter Jost, Crime,
Coordination, and Punishment: An Economic Analysis 21 INT'L REV. L. & ECON. 23 (2001) (allocation of police
resources); David McDowall, Ronald P. Corbett, Jr. & M. Kay Harris, Juvenile Curfew Laws And Their
Influence On Crime 64 FED. PROBATION 58, 60 (2000) (curfew laws); Linda S. Beres & Thomas D. Griffith, Did
"Three Strikes" Cause The Recent Drop In California Crime? An Analysis Of The California Attorney General’s
Report, 32 LOY. L.A. L. REV. 101,111 (1998) (economic opportunity).
59 This assumes that sanctioning costs increase in a linear manner (i.e., if the number of prisoners
increases by 10% so do the sanctioning costs). This assumption is made for analytical simplicity. Also note
that false convictions may carry greater costs than correct convictions in terms of sanctioning costs because
of, for example, the psychological harm caused to innocents who are wrongfully punished.
60 Such a result seems implausible because if deterrence does improve that means that fewer crimes
are occurring which should lead to fewer cases being brought and hence a smaller scope for false
convictions as well. In the extreme if deterrence becomes perfect under a preponderance standard (i.e., zero
crime) then there would be no false convictions, because there are no suits as there are no crimes. Under a
reasonable doubt standard if crimes occur then there is still some chance of a false conviction. See Louis
Kaplow & Steven Shavell, Fairness Versus Welfare, 114 HARV. L. REV. 961, 1244 – 1245 (2001).
61 See Donohue & Siegelman, supra note 54, at 13 (surveying literature, which tends to suggest a .15
elasticity); Steven Levitt, The Effect of Prison Population Size on Crime Rates: Evidence from Prison Overcrowding
Litigation, 111 Q. J. ECON. 319 (1996)(finding higher elasticity figure of .30).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 15
reduction in crime somewhere between .15 and .3 percent. Using the lower
figure, an increase in sanctioning costs of 165% (i.e., tripling the prison
population and increasing sanctioning costs from $71 to $188 billion), should be
associated with a reduction in crime on the order of 25%, ((165)(.15) = 24.75).62
This yields a deterrence benefit of $292 billion. In other words, using as a basis
for prediction the low crime-elasticity estimate of .15, tripling the prison
population should result in a greater than 10% reduction in crime. Since the
sanctioning cost of tripling the prison population ($117 billion) is less than the
expected deterrence benefit ($292 billion), a switch to the preponderance
standard would be desirable on traditional error-cost grounds. If we use the
higher elasticity figure (.3), the benefit from crime reduction would be $582
billion (i.e., a 49.5% decrease in the cost of crime) and it would clearly be
desirable to switch to the preponderance standard. In short, the empirical
evidence suggests that it is very unlikely that increases in sanctioning costs
would fully offset the crime-reduction benefits from switching to the
preponderance rule.63
We thus conclude that it is highly unlikely that the traditional error-cost
rationale for the reasonable-doubt standard is a complete explanation for it. The
expected costs of false convictions generated under the preponderance standard
do not appear likely to exceed the expected costs of false acquittals generated
under the reasonable-doubt rule by a sufficient amount. Put another way,
Blackstone’s assertion that it is better to let 10 guilty men go free rather than
punish one innocent is not supported by a traditional error-cost analysis. Some
other justification must be offered for the pro-defendant bias in the standard of
proof. 64
62 There is a net gain of $175 Billion on these numbers by switching to the preponderance standard.
Crime has dropped by $292 Billion and sanctioning costs have risen to $188 Billion from $71 Billion (i.e., an
increase of $117 Billion). Thus, $292 Billion less $117 Billion is $175 Billion, which makes switching to the
preponderance rule desirable on traditional error-cost grounds.
63 One might ask how much might crime decrease from a switch to the preponderance rule? We
cannot know for sure, but we should note that if we are switching from a 95% standard to a 51% standard
we are substantially increasing the chance that the defendant will be convicted if brought to trial. One
might then surmise that a greater than trivial increase in deterrence is likely to result.
64 In addition, even without the foregoing analysis one might be somewhat skeptical about
traditional error costs explaining the pro-defendant bias in criminal procedure. If one was only interested in
reducing false convictions relative to false acquittals it might be more direct to simply increase the standard
of proof to something beyond reasonable doubt (e.g., virtual certainty) rather to rely on the reasonable
doubt standard plus other protections. Further, Double Jeopardy and the right to a jury trial and so many
other protections may not have particularly clear or significant influences on the number of false convictions
or false acquittals. See Khanna, supra note 35; Timothy Feddersen & Wolfgang Pesendorfer, Convicting the
Innocent: The Inferiority of Unanimous Jury Verdicts (Draft, 1997)(on file with authors)(arguing that under
plausible assumptions the unanimity requirement may result in an increase in false convictions relative to a
supermajority vote requirement). The mere presence of different criminal procedural protections that do not
directly or significantly change error rates suggests that either a more nuanced view of error analysis is
needed or that perhaps error analysis is not a complete explanation for the presence of criminal procedural
protections. See Stuntz, supra note 14, at 1 (suggesting that the procedures are not necessarily correlated with
Toward an Economic Theory of Pro-Defendant Criminal Procedure 16
IV. ANOTHER JUSTIFICATION FOR PRO-DEFENDANT PROCEDURAL PROTECTIONS
Even if the traditional justification for the reasonable-doubt standard is
not altogether satisfying or complete, perhaps, other justifications may provide
us with greater support for it. One such justification, we suggest, is that the
reasonable-doubt standard is designed to make it harder for individuals and
groups to use the criminal process as a mechanism for wealth extraction (or to
obtain their own desired ends). By making the criminal process harder to abuse
we make it less likely that people or certain political groups will want to spend
the resources to lobby prosecutors to enforce the law in certain ways. This
reduces the social costs associated with lobbying and corruption, and enhances
the deterrent effect of the law.
In explaining this justification for pro-defendant criminal procedure it is
important to step back and examine what prosecutors are supposed to do and
how we might attempt to ensure that they achieve those goals efficaciously. The
first point to note is that prosecutors are agents of society in terms of enforcing
the law. As agents it is possible that prosecutors’ incentives may sometimes
diverge from those of society (i.e., from maximizing social welfare). In other
words, prosecutors’ self-interest may not always match up with society’s
interests. This might occur for a number of reasons.
First, prosecutors might not value the same things that society might. For
example, society might be trying to maximize the number of correct convictions
and minimize the number of false convictions subject to a budget constraint.65
On the other hand, prosecutors might be interested in trying to maximize the
number of convictions, to advance further in their careers, make money, or a
variety of other things.66 As prosecutors and society may be maximizing
different things we might expect some divergence between the prosecutor’s
guilt or innocence). This suggests that the traditional error cost justification seems a weak justification. See
Stuntz, supra, at 100.
65 See Khanna, supra note 35. For a similar approach, also see Edward Glaeser & Andrei Schleifer,
Incentives For Enforcement, Draft 2000; Scott Baker & Claudio Mezzetti, Prosecutorial Resources, Plea
Bargaining, and the Decision to Go To Trial, forthcoming 17 J. L. ECON. & ORG’N (2001) (using a model where
prosecutors are concerned about correct convictions and false convictions amongst other things); Dirk G.
Christensen, Comments: Incentives vs. Nonpartisanship: The Prosecutorial Dilemma in an Adversary System, 1981
Duke. L. J. 311 (1981); Darryl K. Brown, Criminal Procedure Entitlements, Professionalism, and Lawyering Norms,
61 OHIO ST. L.J. 801, 801 (2000).
66 See Edward L. Glaeser et al., What Do Prosecutors Maximize? An Analysis of the Federalization of
Drug Crimes, 2 J.L. & ECON. 259, 262-266 (2000); Gordon Van Kessel, Adversary Excess in the America; Criminal
Trial, 67 NOTRE DAME L. REV. 403, 441 (1992); William M. Landes, Economic Analysis of the Court, 14 J. LAW &
ECON. 61 (1971); Standen, supra note 16, at 1477-78; Daniel C. Richman, Essay Old Chief v. United States:
Stipulating Away Prosecutorial Accountability?, 83 VA. L. REV. 939, fn. 93 (1997); Christensen, supra note 92, at
321.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 17
behavior and what is socially desirable behavior. 67 This is one source of agency
costs.
Second, even if prosecutors and society value the same things, each may
weigh them differently. For example, society might value the avoidance of a
false conviction more than a prosecutor might. Society might also value certain
convictions more than prosecutors might (e.g., society might value the conviction
of one drug overlord more than 20 convictions of small time drug dealers, but
prosecutors might view things differently).68
Third, error by prosecutors in assessing the social value of certain cases
could be important. For example, prosecutors might value the same things as
society and, in theory, give them the same weight as society might, but may have
imperfect information so that results congruent with societal interests may not
occur due to error.69 The presence of these errors further widens the gap
between prosecutorial behavior and behavior that maximizes social welfare.
Because prosecutorial interests may not match societal interests in all cases (or
prosecutors may make errors) we face an agency cost problem (prosecutors are
agents for society with interests and behavior divergent, at times, from the social
ideal). This can manifest itself in certain behaviors that might induce lobbying of
prosecutors.
Lobbying prosecutors to bring cases selectively (a type of rent-seeking)
can occur in a variety of forms. However, we think two general types capture
the observed forms of rent-seeking. One is inter-group wealth expropriation, which
67 Of course, if society/the government could costlessly draft an agreement with prosecutors that
perfectly aligned prosecutors’ and society’s interest then there would be little agency problem. As this
seems unlikely, we are faced with diverging interests and agency costs. Cf. Christensen, supra note 65, at
311. This is analogous to the gap between private and social incentives to sue. See Steven Shavell, The
Fundamental Divergence Between the Private and the Social Motive to Use the Legal System, 26 J. Law & Econ. 575
(1997).
68 Note this seems to depend on how (and for what) prosecutors are rewarded. See Christensen,
supra note 65, at 311; Daniel C. Richman, Federal Criminal Law, Congressional Delegation and Enforcement
Discretion, 46 UCLA L. REV. 757, 818 n. 101 (1999); Tracey L. Meares, Rewards for Good Behavior: Influencing
Prosecutorial Discretion and Conduct with Financial Incentives, 64 FORDHAM L. REV. 851 (1995). See Glaeser, supra
note 66, at 261 (noting that decisions to prosecute are often influenced by prosecutors’ interest in running for
a higher office); Van Kessel, supra note 66, at 442 (1992) (noting how win-loss records are very important to
state prosecutors). This raises interesting questions about why we reward prosecutors in the way we do.
This is the subject of a separate paper and outside the scope of our present inquiry. As a first cut, one
suspects convictions rates are the more verifiable assessment criteria relative to others (much like profits are
a bit easier to verify as assessment criteria relative to others in the corporate sphere when dealing with the
agency problem in that context).
69 For example, assume that the prosecutor does value the same things and to the same degree as
society. Thus, the prosecutor wants to maximize correct convictions, minimize false convictions subject to a
budget constraint and values the avoidance of a false conviction as much as society. However, the
prosecutor may incorrectly believe that a certain defendant is guilty even when in fact he may not be. In
this situation, the prosecutor may pursue this innocent defendant even though the prosecutor values the
same things and to the same extent as society.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 18
arises when one group attempts to gain some advantage from the prosecutor at
the expense of other members of society.70 Equivalently, we might describe this
type of rent-seeking/lobbying as an advanced version of tribalism. We treat this
as wealth expropriation – group A attempts to obtain gains at the expense of
other groups in society (we will refer to all non-A groups as group B for
simplicity). The other general type of rent-seeking is simple corruption, which
occurs when an offender or potential offender uses bribery or some other means
to induce an enforcement agent to selectively enforce the law. Using these
conceptions of rent-seeking we can discuss some examples.
A. Inter-Group Wealth Expropriation
Rent-seeking behavior is a result of prosecutors acting in their
unconstrained self-interest. If there are no constraints prosecutors may use the
criminal process to benefit themselves by selling, in some sense, their ability to
enforce the criminal law to the highest bidder. What prosecutors receive could
include direct or indirect monetary gain, enhanced chances for power and
prestige, or anything else of value to the prosecutor.71 To simplify we can say
that prosecutors receive a certain sum – say $1 million – from the highest bidding
group (A) to enforce the law in a particular manner. Enforcement of the law in
this manner must benefit the highest bidding group by more than $1 million –
say $1.5 million – and these gains would appear to come at the expense of the
non-A groups (i.e., group B for simplicity). Thus, in a sense, A is using the
prosecutor and the criminal process to extract $1.5 million from B by paying the
prosecutor $1 million. The various groups realizing that the prosecutor is willing
to sell his services will lobby to secure some of these gains from prosecutors and
to prevent other groups from extracting wealth from them.72 The lobbying and
counter-lobbying efforts, we will see below, can generate significant social
costs.73
70 See Einer R. Elhauge, Does Interest Group Theory Justify More Intrusive Judicial Review?, 101 YALE
L.J. 31, 35 (1991); Richard H. McAdams, Cooperation and Conflict: The Economics Of Group Status Production
And Race Discrimination, 108 HARV. L. REV. 1003, 1029-1030 (1995) (noting that “[W]hen interest groups
pursue what economists call "rent-seeking" legislation, such as farm subsidies and tax "loopholes," they seek
merely to transfer resources from one group to another”); Angela O. Davis, The American
Prosecutor:Independence, Power, and the Threat of Tyranny, 86 IOWA L. REV. 393, 459 (2001).
71 See, e.g., Richman, supra note 68, at 818, n. 102 (noting prosecutors’ eagerness for career
advancement); Christensen, supra note 65, at 318 (noting that prosecutors are financially motivated).
72 See MUELLER, infra note 92. See also Shavell, supra note 67, at 612 n. 46 (1997)(discussing factors
that influence prosecutors from state compensation to collective private efforts).
73 See Paul H. Rubin & Martin J. Bailey, The Role of Lawyers in Changing the Law, 23 J. LEGAL STUD.
807, 822-23 (1994)(discussing how rent-seeking motives of lawyers result in social waste). We have assumed
that each group acts as a monolith. That is, we are abstracting away from intra-group sharing issues. Of
course, in reality even within each group there may be some competition for the rents that the group earns.
Discussion about how this affects our analysis is left for another time. See generally Gary S. Becker, A Theory
of Competition Among Pressure Groups for Political Influence, 98 QUARTERLY JOURNAL OF ECONOMICS 371 (1983);
Sun & Ng, infra note 104 (noting how size and number of interest groups affects rent-dissipation (i.e., may
initially increase then decrease)).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 19
Inter-group wealth expropriation can be effected in a number of different
ways. For example, there is lobbying that results in targeted enforcement. This
occurs when prosecutors disproportionately target certain groups in society for
purposes of bringing prosecutions due to the lobbying efforts of certain other
groups. The quintessential case would be where prosecutors disproportionately
brought charges against members of group B because of the lobbying efforts, or
simply to curry the favor, of the dominant group A. This permits group A
members to shift the burden of criminal enforcement they might otherwise bear
onto group B members or to otherwise impose costs on group B members which
result in some kind of benefit to A (e.g., the maintenance of a caste system). 74
Indeed, in regimes in which prosecutors are elected, candidates for the position
will have incentives to seek support from group A members by promising to
direct enforcement efforts against group B members.75 Perhaps the best known
example of this in United States history is law enforcement in the South during
the Jim Crow period, which involved numerous instances of prosecutors
refusing to enforce the law against white citizens, while using the threat of
criminal punishment to coerce black citizens.76
Another example is where prosecutors directly ask for some benefit to
avoid bringing charges against members of politically marginal groups. This is
akin to extortion or a protection racket. Knowing that politically powerful
groups will have him removed from office if he threatens their interests, the self-
interested prosecutor could focus his extraction efforts on the politically
marginal.77 In this version of wealth expropriation, the prosecutor seeks to
extract wealth from members of politically marginal groups, while providing
benefits to the politically powerful in order to keep his position. The difference
between this version of inter-group wealth extraction and the first is slight: in the
first, the powerful group initiates the wealth extraction process and in the
74 See RICHARD EPSTEIN, FORBIDDEN GROUNDS 91-97 (1992); Dwight L. Greene, Symposium: Criminal
Law, Criminal Justice and Race Justice Scalia and Tonto, Judicial Pluralistic Ignorance, and the Myth of Colorless
Individualism in Bostick v. Florida, 67 TUL. L. REV. 1979, 1982 (1993); Kathleen M. Sullivan, Unconstitutional
Conditions, 102 HARV. L. REV. 1413, 1498 (1989).
75 See Daniel C. Richman, supra note 68, Stuntz, supra note 10, at 20 – 37 (discussing the incentives of
various participants in the American Criminal Justice system); John A. Horowitz, Prosecutorial Discretion and
the Death Penalty: Creating a Committee to Decide Whether to Seek the Death Penalty, 65 FORDHAM L. REV. 2571
(1997); Tracey L. McCain, The Interplay of Editorial and Prosecutorial Discretion in the Perpetuation of Racism in
the Criminal Justice System, 25 COLUM. J.L. & SOC. PROBS. 601, 648, n. 81 (1993) (noting that decisions to
prosecute are susceptible to political influence because most prosecutors are elected); Dwight L. Greene,
Abusive Prosecutors: Gender, Race & Class Discretion and the Prosecution of Drug-Addicted Mothers, 39 BUFF. L.
REV. 737, 777 (1991) (noting that “Prosecutors … are capable of conducting their offices in ways to advance
their own political careers”).
76 See EPSTEIN, supra note 74; William J. Stuntz, Race, Class, and Drugs, 98 COLUM. L. REV. 1795, 1839
(1998) (noting that robbery laws in the Jim Crow era were enforced against blacks more often than whites
especially where white robbers stole from black victims).
77 See William N. Eskridge, Religion, Homosexuality, and Collisions of Liberty and Equality in American
Public Law, 106 YALE L.J. 2411, 2447 (1997); Greene, supra note 75, at 799.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 20
second, the prosecutor initiates the extraction process. In the first case, the
powerful group is likely to pay the prosecutor the smallest amount necessary to
accomplish their ends, allocating the surplus to themselves. In the second, the
prosecutor who initiates the wealth extraction process will allocate the surplus to
himself. One example of this process occurs in rural areas of China, where local
police officers have tried to enrich themselves by enforcing certain prohibitions,
such as the one-child policy, against relatively poor farmers.78
Yet another example is the passing of laws (including criminal laws) with
disproportionate burdens on different groups.79 This is analogous to targeted
(i.e., disproportionate) enforcement. In order to effect wealth-expropriation, the
laws need not apply directly to group B members. The dominant group (A) may
find that certain activities are carried out only, or predominantly, by group B
members, or that group B members carry out these activities in a different
manner from others. 80 With this information, the dominant group may prohibit
or place special burdens on the activity, or the activity when carried out in a
particular manner. For example, white majorities in the western United States
enacted several facially neutral statutes in the late 1800s that had the effect of
prohibiting Chinese laundries, both to limit competition from them and to limit
the independent work options of Chinese laborers.81
All three of these cases are united by a common theme: one group benefits
at the expense of others – wealth expropriation – by using the governmental
process, whether law enforcement, legislation, or adjudication. As our concern is
with prosecutorial behavior, we will not discuss in much depth the passing of
laws with disproportionate burdens as that is the legislative context.82 However,
it is important to note that the legislative and law enforcement processes provide
alternative routes through which a predatory dominant group could extract
78 See Elisabeth Rosenthal, Rural Flouting of China’s One-Child Policy Undercuts Census, N.Y. TIMES,
April 14, 2000.
79 See, e.g., Stuntz, supra note 76, at 1795 (discussing the heightened police attention given to urban
crack markets dominated by the lower economic class while law enforcement pays relatively little attention
to the upscale powder-cocaine market); 1976 Supreme Court, Term 1: Constitutional Significance of Racially
Disproportionate Impact, 90 HARV. L. REV. 114, 119 (1976).
80 See Robert C. Post & Reva B. Siegel, Equal Protection by Law: Federal Antidiscrimination Legislation
After Morrison and Kimel, 110 YALE L.J. 441, 469, n. 149 (2000) (noting the disparate impact on women that the
facially neutral marital rape exceptions had); Kevin R. Johnson & George A. Mason, Discrimination By Proxy:
The Case of Proposition 227 and the Ban on Bilingual Education, 33 U.C. DAVIS L. REV. 1227, 1245 (2000) (noting
that “tough on crime” laws such as the “three strikes” laws, have “racial bents” although written in neutral
terms).
81 See David E. Bernstein, Lochner, Parity and the Chinese Laundry Cases, 41 WM. & MARY L. REV. 211
(1999); Stacy Suzan Kahana, Crossing the Border of Plenary Power: the Viability of an Equal Protection Challenge to
Title IV of the Welfare Law, 39 ARIZ. L. REV. 1421, 1436 (1997).
82 See generally, William E. Adams, Is it Animus or a Difference of Opinion? The Problems Caused By the
Invidious Intent to Anti-Gay Ballot Measures, 34 WILLIAMETTE L. REV. 449 (1998) (discussing the legislative
measures aimed at homosexuals); Kahana, supra note 81, at 1421.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 21
wealth from other groups. 83 In this sense, the legislative and law enforcement
processes are substitutes in the eyes of the wealth extractor. The laws controlling
lobbying/rent-seeking in these governmental processes can often be understood
as complements, in the sense that they prevent a predatory dominant group from
shifting its expropriation efforts from one governmental process to another.84
B. Simple Corruption
The other type of rent-seeking, simple corruption, involves the effort of a
single individual or group to extract wealth from the general population. We
have in mind two cases: that of an enforcement agent (police officer or
prosecutor) who threatens to apprehend and charge an individual unless he pays
the agent (e.g., extortion), and that of an enforcement agent who is merely
willing to accept bribes from the general public.85 In general, these payoffs can
take two forms. One, ex ante bribery, occurs when an individual bribes an
enforcement agent before he commits a crime in exchange for an agreement by
the agent not to enforce the law against him. In the other form, ex post bribery,
the individual bribes the agent after he commits the crime.86
There are numerous examples of simple corruption. A common example
of ex post bribery is a police officer that accepts bribes in return for not issuing a
ticket to a speeding motorist. Ex ante bribery appears to be less common, though
there are many examples of it too. In most towns in the U.S., local government
business is carried out by boards made up of residents with deep and strong
connections to many of the parties who appear before them.87 In these settings, it
is hard to distinguish the ordinary reciprocal exchanges that are part of normal
social intercourse from ex ante bribery. The Supreme Court grappled with a
rather routine example of this in City of Columbia v. Omni Outdoor Advertising,88
which involved the efforts of a local billboard company to protect its incumbency
83 See Frank B. Cross, The Role of Lawyers in Positive Theories of Doctrinal Evolution, 45 EMORY L.J. 523
(1996).
84 For discussion of how the procedural protections may induce greater lobbying at the legislative
level see Stuntz, supra note 10 passim. We discuss this matter at infra text accompanying notes 157 – 160.
85 See James Lindgren, The Theory, History, and Practice of the Bribery-Extortion Distinction, 141 U. PA.
L. REV. 1695, 1701 (1997). One could argue that the simple corruption category is the same as our second
example of inter-group expropriation, and we concede that the difference is more a matter of degree than of
character. In the second example of wealth expropriation, the enforcement agent maintains his position
through the support of local dominant groups. In the simple corruption story, the enforcement agent is
either unconcerned with maintaining support from local dominant groups (in the case of the actively
predatory enforcer), or passively accepts bribes in exchange for not enforcing the law.
86 See Mehmet Bac & Parimal Kanti Bag, Law Enforcement Costs and Legal Presumptions 5 - 6 (Draft,
2000)(on file with authors). See also Mehmet Bac, Corruption, Supervision and the Structure of Hierarchies, 12
J.L. ECON & ORG’N. 277 (1996).
87 See generally Mehmet Bac, The Scope, Timing and Type of Corruption, 18 INT’L REV. L. & ECON. 101,
104 (1998).
88 499 U.S. 365 (1991).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 22
advantage by encouraging the city council to prohibit the erection of new
billboards. 89
Thus, because of the gap between prosecutorial and societal interests,
there are agency costs that are manifested in the behavior of parties seeking to
expropriate wealth through lobbying for selective enforcement of the law. In
light of this it becomes important to get a sense of the costs generated by such
behavior (Part V) and also some methods for constraining this kind of behavior
(Part VI). 90
V. COSTS ASSOCIATED WITH “RENT-SEEKING” BEHAVIOR IN THE CRIMINAL
PROCESS
We divide the discussion of the costs of lobbying or rent-seeking into two
parts. First, in section A, we discuss the costs related to the act of lobbying.
Second, in section B, we discuss the costs related to the effect of lobbying (and
the perception of successful lobbying) on the deterrent force of the criminal law.
A. Direct Costs From Rent-Seeking – Wasteful Expenditures
Rent-seeking in the criminal process involves lobbying efforts by certain
individuals or groups to influence the selection and prosecution of cases. The
process of lobbying itself generates costs that are, from a societal perspective,
often essentially wasteful. In order to discuss these costs in greater depth a
useful starting point is an analogy to the efforts to obtain a monopoly. Monopoly
status provides the person holding it with the ability to extract supra-competitive
prices for some period of time and hence make supra-normal profits.91 These
profits are attractive and are likely to induce people to spend resources on
obtaining this monopoly. 92 This expenditure of resources is sometimes socially
desirable and at other times socially undesirable.
Expenditures to obtain a monopoly may be desirable when a firm secures
a dominant position through competition because a firm typically does this by
improving its product, or reducing its costs, activities that increase the total
89 The excluded firm brought an unsuccessful antitrust lawsuit on the ground that the incumbent
firm had colluded with city officials. The Supreme Court’s reluctance to apply the antitrust laws to this
behavior is based in large part on the difficulty in distinguishing ex ante bribery from ordinary social
intercourse. See id. at 379 – 80.
90 This is similar to the conflict of interest that can be a problem for individuals with decision-
making authority in a corporation since these individuals may have to decide between their own welfare
and the welfare of the corporation. See ROBERT CHARLES CLARK, CORPORATE LAW 147 (Little, Brown, &
Company 1986).
91 See PHILLIP AREEDA & LOUIS KAPLOW, ANTITRUST ANALYSIS: PROBLEMS, TEXT, AND CASES 16 (5th ed.
1997).
92 See DENNIS C. MUELLER, PUBLIC CHOICE II 229 – 46 (rev. ed. 1989)(1979).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 23
surplus.93 In general, this is a benefit to society because competition to be the
“best” inures to the benefit of consumers and society. 94
However, certain expenditures to obtain or maintain a monopoly are
wasteful from society’s perspective. Such expenditures include duplicative
lobbying efforts to obtain a government privilege (e.g., an exclusive license or
tariff protection), to obtain certain kinds of governmental behavior, to retard
research progress, and others.95 This kind of naked wealth extraction creates no
additional surplus (i.e., no benefit to society) and merely transfers an asset from
one party to another.96 The gap between the resources expended and the benefit
to society (the benefit is zero in the case of pure transfer of wealth) is wasteful
from society’s perspective.97
In the context of criminal law enforcement, efforts to lobby the prosecutor
are often wasteful in a sense similar to naked wealth extraction. If we assume an
unbiased prosecutor then lobbying such a prosecutor to bring selective
enforcement against one group (say, group B) by members of group A could be
wasteful in certain instances. Of course, the result is not wasteful if targeting
group B reduces the overall costs of crime.98 However, there is little reason to
believe that lobbying for selective enforcement when there is an unbiased
prosecutor will always bring about an efficient result.99 For example, group A
will have no interest in inducing the prosecutor to go after cases of crime
involving only members of group B. Further, group A members may discourage
the prosecutor from enforcing the law when members of their own group
93 See AREEDA & KAPLOW, supra note 91, at 7 (noting that “[c]ompetitive forces generate efficiency in
two ways. Productive efficiency occurs as low cost producers undersell and thereby displace the less
efficient. Allocative efficiency occurs as exchanges in the marketplace direct production away from goods
and services that consumers value less and toward those they value more...”).
94 See id.
95 See MUELLER, supra note 92, at 229 – 46.
96 See id., at 231.
97 See id., at 229-246.
98 It might be that lobbying coincides with what might be socially desirable. Also lobbying is not
necessarily limited to inter-group lobbying. For example, if the victims of crimes by members of group B are
other group B members then group B may lobby prosecutors to stop crime in their areas and hence lobby for
prosecutions against other group B members. This sort of lobbying does not raise the kinds of concerns we
are discussing in this paper. See Gordon Tullock, Efficient Rent-Seeking, in TOWARD A THEORY OF THE RENT-
SEEKING SOCIETY 97 – 112 (James M. Buchanan, R.D. Tollison & Gordon Tullock eds., 1980).
99 Note that even if the prosecutor is unbiased she may still not prosecute the cases that would be in
society’s interests. This stems from the prosecutorial agency costs discussed earlier. See supra Part IV.
Further, lobbying may in some instances overcome (or reduce) the agency problem, but in others it might
exacerbate them. Of course, if the prosecutor has the same interests as society and places the same weights
on things as society does then lobbying that prosecutor can only be efficiency-enhancing if the prosecutor
may make errors in deciding which cases to bring. However, once again, there is no particular reason to
believe that lobbying will generally result in more accurate, as opposed to less accurate, decisions.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 24
commit crimes against group B members.100 In these cases lobbying is a social
waste.101
This waste includes some portions of the lobbying efforts of politically
dominant groups, the counter-lobbying efforts of politically non-dominant
groups, and some portion of the effort and time spent by government officials in
addressing wealth transfers and maneuvering to obtain positions in which they
can direct such transfers.102 The easier it is to obtain governmental favors the
more lobbying should appear.103 Also, as the value of the issues at stake
increases one should expect greater expenditures as well.104
To this point we have largely focused on the costs associated with inter-
group expropriation efforts, but analogous arguments apply in the context of
simple corruption.105 Corruption creates costs in terms of the resources spent in
100 Group A members may do this for a variety of reasons. For example, if they have their own
methods of social control (for A members) besides official law enforcement then they may prefer to rely on
those methods rather than official law enforcement. See Kelly D. Hine, Vigilantism Revisited: An Economic
Model of the Law of Extra-Judicial Self Help or Why Can’t Dick Shoot Henry for Stealing Janet’s Truck?, 47 AM. U.
L. REV. 1221 (1998).
101 Even if lobbying did produce an efficient rule (i.e., an overall reduction of crime) it might be that
there was too much lobbying to achieve that end or that an alternative means of influencing prosecutorial
behavior would have lower costs. See generally GORDON TULLOCK, THE ECONOMICS OF SPECIAL PRIVILEGE AND
RENT SEEKING 11-27 (1989).
102 See MUELLER, supra note 92, at 334 (citing G.S. Becker, A Theory of Competition among Pressure
Groups for Political Influence, QUARTERLY JOURNAL OF ECONOMICS 371, 373-374 (August 1983)); See generally
Jack M. Beermann, Interest Group Politics and Judicial Behavior: Macey’s Public Choice, 67 NOTRE DAME L. REV.
183, 201- 04 (1991).
103 See MUELLER, supra note 92, at 334; Beermann, supra note 102, at 183.
104 See Daniel Fischel & Alan O. Sykes, Corporate Crime, 25 J. LEGAL STUD. 319, 324 (1996); Arthur B.
Laby, W. Hardy Callcott, Patterns of SEC Enforcement Under the 1990 Remedies Act: the Civil Money Penalties, 58
ALB. L. REV. 5, 50 (1994) (noting that more defendants are willing to invest in litigation against the SEC
rather than settle since the penalties have been increased under the Remedies Act); Stephen J. Spurr, An
Economic Analysis of Collateral Estoppel, 11 INT’L REV. L. & ECON. 47 (1991). Also, as the number of groups,
political or otherwise, increase (e.g., the more heterogeneous the population) the more potential lobbying
groups and the more likely that wasteful spending might occur. See also Guang-Zhen Sun & Yew-Kwang
Ng, The Effect of the Number and Size of Interest Groups on Social Rent Dissipation (Draft)(on file with authors),
(suggesting that the amount of resources expended in rent-seeking would probably first increase as the
number of groups increased and then, after some point, begin to taper off).
105 Note that a bribe might be efficient in some cases, much as lobbying for targeted enforcement
might be. See Gary S. Becker & George J. Stigler, Law Enforcement, Malfeasance and Compensation of Enforcers, 3
J. LEGAL STUD. 1 (1974); Padideh Ala’i, The Legacy of Geographical Morality and Colonialism: A Historical
Assessment of the Current Crusade Against Corruption, 33 VAND. J. TRANSNAT’L 877, 899 (2000). Such a situation
would be subject to the same arguments as with targeted lobbying (if efficient). See supra note 99. See also
Rebecca A Pinto, The Public Interest and Private Financing of Criminal Prosecutions, 77 WASH. U. L. Q. 1343,
1367 (1999) (noting that “private financing has the potential to further public values by providing a
corrective offset of fiscal and other institutional influences on prosecutorial discretion that lead to
inequitable allocation of criminal justice resources. Canceling out such influences would enable prosecutors
to make charging decisions on a financially-level playing field and enhance the prosecutor’s freedom to
pursue cases and offenders most deserving of prosecution.”); See also Tracey L. Meares, Rewards for Good
Behavior: Influencing Prosecutorial Discretion and Conduct with Financial Incentives, 64 FORDHAM L. REV. 851
(1995).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 25
bribing the enforcement agent and the efforts of the agent in positioning himself
to receive bribes. However, there is an important feature of the corruption
model that suggests that rent-seeking costs can be much larger than appears
initially. The enforcement process is vertically fragmented, in the sense that it
moves through a chain beginning with a police officer (and perhaps next his
superiors), moving to a prosecutor, and on to a magistrate or judge, and so on. If
each one of these agents demands bribes, then the total social waste will be
considerably larger than in a vertically integrated enforcement regime in which a
single agent controls the process from arrest to punishment.106
The vertical fragmentation of the enforcement process means that each
individual agent is in a position similar to that of successive owners of the pieces
of a long toll road. One of the standard results of economics is that sum of the
tolls charged by successive owners will be larger than the toll charged by a single
owner of a road.107 The reason is that successive owners do not take into account
the fact that any increase in their individual tolls will reduce the revenue to the
owners of the other pieces of the toll road. The same phenomenon is likely to be
observed in the law enforcement process when corruption is rampant.
The problem here is entirely analogous to the familiar double-marginalization problem in the
theory of vertical integration. See Andy C.M. Chen & Keith N. Hylton, Procompetitive Theories of Vertical
Control, 50 Hastings L.J. 573, 623 (1999) (noting that “the monopolist manufacturer must worry about the
‘double-marginalization’ problem…the monopolist will apply a monopoly surcharge to the item he
produces, and the distributor will apply an additional monopolistic surcharge to the same item at the
downstream level.”). A fragmented enforcement process allows several individuals to impose a monopoly
surcharge for their services. A vertically-integrated enforcement process allows one individual to impose a
surcharge. See id. at 624 (noting again that “the monopolist can eliminate this problem by vertically
integrating forward, and transferring his own products at marginal cost to the downstream segment of the
integrated unit.”). The overall costs of corruption and reduction in service are considerably greater under
the fragmented regime. See Schleifer & Vishny, infra note 134.
106 We are not necessarily suggesting that law enforcement should be vertically integrated (this
may raise “checks and balances” concerns), but we are saying that fragmenting enforcement can increase the
amount of loss from corruption.
107 See Howard A. Shelanski & J. Gregory Sidak, Antitrust Divestiture in Network Industries, 69 U.
CHI. L. REV. 1, 23 (2001) (arguing that “under simple models of vertical control where the downstream firm
is assumed to have market power, social welfare is unambiguously increased by the elimination of the
double marginalization. One firm rather than two marks up the price of the upstream product, leading to a
lower price and higher output.”); See John E. Lopatka & Andrew N. Kleit, The Mystery of Lorain Journal and
the Quest for Foreclosure in Antitrust, 73 TEX. L. REV. 1255, 1297 (1995) (arguing that “if the products were
complements, their provision by a single supplier could increase efficiency, for monopoly provision of
complementary products can avoid the kind of double-marginalization problem well recognized in the
context of vertically related, or successive, monopolies. In essence, when complementary inputs are
supplied by different monopolists, price may be higher and output lower than if both are supplied by a
single seller.”) For discussion of a related problem – the problem of the anti-commons see Michael A.
Heller, The Tragedy of the Anticommons: Property in the Transition from Marx to Markets’, 111 HARV. L. REV. 621
(1998); James Buchanan & Yong J. Yoon, Symmetric Tragedies: Commons and Anticommons Property,
Journal of Law and Economics, (forthcoming 2000); Francesco Parisi, Ben Depoorter, & Norbert Schulz, Duality
in Property: Commons and Anticommons, University of Virginia School of Law, Law & Economics Research
Papers Series, Research Paper No. 00-16 (Draft 2001) (on file with authors).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 26
Although the costs of lobbying and corruption might be quite large that
does not end the potential costs associated with this kind of behavior. In the
context of the criminal process such lobbying, corruption, and general rent-
seeking (and in particular the perception of this kind of behavior) might have
effects on compliance with the law and with deterrence that need to be
accounted for as well.
B. Deterrence and Other Costs from Rent-Seeking
In the context of the criminal process rent-seeking (if successful in any
measure) has the effect of skewing the enforcement of the law. If members of
society perceive (i) skewed enforcement (i.e., disproportionate enforcement) and
(ii) that such enforcement is related to rent-seeking/lobbying efforts by certain
groups (i.e., that law enforcement is “political”) then compliance with the law
and the deterrent effect of the law are likely to be compromised. The effects on
compliance and deterrence are multi-faceted and we address some of them in the
following sections.
1. Some deterrent effects of selectively enforcing the law
Selective enforcement of the law due to lobbying by certain groups can
have corrosive effects on deterrence. In this section we consider two scenarios.
First, where lobbying by group A members results in selective law enforcement
against group B members with little regard to the actual guilt of the defendants
and disproportionately less enforcement against group A members (case 1).
Second, where lobbying by group A members leads to selective enforcement
against group B members focusing on those who are guilty, but again with
disproportionately less enforcement against group A members (case 2).
In case 1, deterrence is likely to drop for both groups A and B. The reason
is as follows. Deterrence can be achieved through substitution effects or scale
effects. Substitution effects occur when a change in the effective sanction leads
potential offenders to substitute legitimate, law-complying conduct for
illegitimate, undesirable conduct. Scale effects occur when enforcement causes
potential offenders to stay out of certain areas, or off the streets at certain times.
A selective enforcement policy in which group B is targeted implies, within a
fixed budget setting, a diversion of resources from substitution-oriented policies
to scale-oriented policies. This is analogous to shifting from a strategy of
ticketing every motorist that speeds (inducing substitution toward slow driving)
to a strategy of ticketing motorists who meet the profile of a speeder (inducing
drivers who fit the profile to stay off the roads).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 27
Now, in case 1, the group A members will be under-deterred because they
are facing low expected sanctions for engaging in undesirable activities (as law
enforcement occurs less frequently against them due to their lobbying).108
Consequently, they have an increased incentive to engage in these activities
relative to where law enforcement was not biased in their favor. The deterrent
effect on group B members would also be reduced. This is because they are now
punished whether they have acted “good” or “bad.”109 In other words, the
incentive to act “good” is reduced for group B members because the payoffs
from acting “good” and “bad” have gotten closer. Thus, deterrence for both
groups (A and B) drops.
An enforcement policy chosen by group A members that reduces
deterrence may seem irrational, because it could lead to additional crimes being
committed against group A members. However, there are scenarios in which
such a policy could be chosen rationally by group A. For example, if groups A
and B are geographically segregated, group A may choose to reduce potential
crimes by Bs on As by apprehending all Bs who venture into their territory,
whether or not the Bs are complying with the law.110 Such a policy would make
it costly for Bs to move among As, encouraging the Bs to stay in their own
territory.111 In other words, group A members may choose to rely on scale effects
to reduce the risks posed to them by group B offenders. Such a policy, in a fixed
108 See A. Mitchell Polinsky & Steven Shavell, Legal Error, Litigation, and the Incentive to Obey the Law,
5 J. L.
ECON. & ORG’N 99 (1989) (noting that if a potential defendant will not be prosecuted whether or not he
obeys the law, he “will not obey the law, because there is a cost to obeying but no benefit (assuming, of
course, that there is a benefit in disobeying)”.
109 See id., at 104 (stating that “A type II error (a truly innocent defendant is found liable) lowers the
incentive to obey the law because he will face liability even if he obeys, thereby reducing the benefit to him
of obeying the law”).
110 See, e.g., David A. Harris, The Stories, The Statistics, and the Law: Why “Driving While Black”
Matters, 84 MINN. L. REV. 265, 271(1999)(discussing law enforcement’s tendency to subject black drivers in
upscale neighborhoods to traffic stops); Tracy Meares, Social Organization and Drug Law Enforcement, 35 AM.
CRIM. L. REV. 191, 203-4 (noting that “Law-abiding African Americans who are unable to physically separate
themselves from victimizers might desire to draw legal boundary lines in order to reconstitute healthy, law-
abiding communities around them. These law-abiding African Americans might argue in favor of calling
upon the machinery of the state to make a distinction between themselves and law-breaking African
Americans who reside in their neighborhoods. Under this view, imprisonment might be the most effective
form of line-drawing because imprisonment distinguishes law-abiders from law-breakers by removing law-
breakers from the community. By relying on incarceration, law-abiders can create physical distances
between themselves and law-breakers. In this way, severe legal sanctions that lead to the removal of law-
breakers from the community are akin to leaving the neighborhood. Indeed, this reasoning suggests that
victimized law-abiding African Americans should welcome state enforced distinctions between law-abiding
and law-breaking African Americans”).
111 This is one potential explanation for racial profiling, to the extent it exists. For some empirical
evidence on racial profiling see John Knowles, Nicola Persico & Petra Todd, Racial Bias in Motor Vehicle
Searches: Theory and Evidence, 109 J. POL. ECON. 203 (2001); John J. Donohue III & Steven Levitt, The Impact of
Race on Policing, Arrest Patterns, and Crime (Draft 1998)(on file with authors).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 28
budget setting, could easily result in a weakening at the substitution-effect level
in deterrence among Bs.112
Case 2 presents the same potential under-deterrence problem for group A
members because again they may face low expected sanctions. However, for the
group B members the situation has changed. The deterrent effect on group B
members may still remain or be enhanced because in this case if a group B
member acts “good” he is probably not going to be prosecuted and if he acts
“bad” he is more likely to be prosecuted.113 However, given that A members are
assumed to control the enforcement process, the more probable scenario is one in
which their agent (the prosecutor) is unable, because of unfamiliarity or
indifference, to distinguish complying from non-complying members of group B.
With a prosecutor unable to distinguish the “good” from the “bad” among group
B, a policy of targeting the bad in group B would be infeasible. This suggests
that case 1, in which deterrence clearly falls, is the more likely outcome of
selective enforcement.
2. Other costs: stigma effects, expressive effects, enforcement cost effects
Rent-seeking behavior may also reduce the stigma associated with the
criminal law and thereby dilute deterrence. If the criminal label carries some
stigma then that becomes part of the total sanction (i.e., the official sanction plus
the stigma) for criminal behavior and influences deterrence.114 If something
reduces the stigma without a countervailing increase in the official sanction then
deterrence is reduced because potential wrongdoers face a lower expected
sanction. 115
112 Also, group A members may want to have less enforcement against other group A members for
a variety of reasons including that some group A crime may victimize group B members, group A may have
its own methods of addressing crime by group A members on other group A members, and so forth. In the
case of crime by A on A it is possible that A may prefer to avoid official law enforcement. This could
involve issues of intra-group wealth extraction, other benefits from avoiding enforcement by officials
(reputation, etc…), and that alternative means of policing A members may be available (e.g., social norms).
If the social norms option is the driving force it may be that deterrence in group A would not suffer as a
result of lobbying for less official enforcement because the social norms may engender adherence to the law
or the norm. Of course, this raises all kinds of issues about whether the social norms track the law, are the
norms efficient, and whether the amount of deterrence obtained through social norms is the same as the
amount through official law enforcement. See Jonathan Simon, Law, Democracy, and Society: Megan’s Law:
Crime and Democracy in Later Modern America, 25 LAW & SOC. INQUIRY 111, 1119 (2000) (arguing that “in most
of the original cases charges resulted in acquittals or were never brought at all, as juries, prosecutors, and
police seemed to collude in sheltering concededly fringe elements of the white community”).
113 Since under this scenario, the marginal expected sanction either increases or stays the same,
deterrence (at the substitution-effect level) remains intact.
114 See Khanna, supra note 31, at 1508 – 9; David Cole, The Paradox of Race and Crime: A Comment on
Randall Kennedy’s “Politics of Distinction”, 83 GEO. L. J. 2547, 2561 (1995).
115 See Dan M. Kahan, Social Influence, Social Meaning, and Deterrence, 83 VA. L. REV. 349, 350 (1997);
Richard Moran, Home Sweet Home: Given a Choice, Many Convicts are Now Opting for Jail Instead of Probation:
Why?, BOSTON GLOBE, October 29, 2000, at E1.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 29
One suspects that the stigma from being labeled a criminal stems in part
from a belief that the person so labeled has violated some societal norm meriting
condemnation and has been adjudicated in a “fair and impartial” manner.116 If,
however, being labeled a criminal is perceived by members of your social circle
as partially indicative of a political/biased use of the law, rather than that you
are worthy of condemnation, then the stigma from the criminal label likely
diminishes.117 In this case the stigma from being labeled a criminal is less than
the stigma suffered by someone labeled a “criminal” who is part of a group of
people who believe that adjudication is fair and impartial to their group. 118
This seems consistent with some anecdotal (and empirical) evidence that
sometimes criminal sub-cultures form that do not consider the criminal label to
be terribly stigmatizing (the criminal label might actually enhance one’s respect
amongst certain groups).119 If there is a differential stigma impact when rent-
seeking or lobbying is perceived to be a problem as compared to when it is not
then the presence of rent-seeking reduces the stigma suffered by violators and
hence reduces the deterrent impact of the criminal law. 120
For reasons similar to those related to the stigma argument, greater rent-
seeking is likely to reduce any “expressive” features of the criminal law. The
expressive effects of the criminal law are argued to take many forms and we
focus on two to highlight our argument.121 First, that the criminal law expresses
to society what is undesirable behavior and this encourages individuals not to
engage in this activity because they respect (or otherwise value) the law (i.e.,
norm or law internalization). 122 If so, then a perceived increase in rent-seeking
116 See Paul H. Robinson & John M. Darley, The Utility of Dessert, 91 NW.U. L. REV. 453 (Winter 1997)
(noting the message that being labeled a “criminal” sends to the community); Janice Nadler, The Effects of
Perceived Injustice on Deference to the Law, at 9-10, Draft 2000 (on file with authors).
117 See ERIC POSNER, LAW AND SOCIAL NORMS 97 – 100 (2000) (noting that criminal offenders can
signal loyalty to a subcommunity by violating the law and being punished by the dominant group. The
subcommunity is more likely to view criminal punishment as a signal of loyalty to the subcommunity the
more the subcommunity believes the criminal justice system is “infected with a political agenda”); Dan
Kahan, supra note 115, at 357 – 58; Nadler, supra note 116, at 10 (suggesting that, if the law is seen to be
imposed in an irrelevant or immoral manner, it will not be deferred to).
118 See POSNER, supra note 117, at 98; Kahan, supra note 115; Nadler, supra note 116, at 10. For this
argument to work all we need is that stigma is different (and lower) if some part of your social circle will not
ostracize you because of perceived misuse of the criminal process.
119 See Kahan, supra note 115, at 357 (suggesting that sometimes gang members wear their
convictions and prosecutions like “Badges of Honor”); Moran, supra note 115.
120 This seems to suggest differential sanctions for different groups in society based on how much
stigma they are likely to perceive from the criminal label. However, engaging in differential sanctioning
may further exacerbate the perceived political use/misuse of law enforcement.
121 See Richard H. McAdams, A Focal Point Theory of Expressive Law, 86 VA. L.REV. 1649, 1659 – 63
(2000)(citing THOMAS C. SCHELLING, THE STRATEGY OF CONFLICT 54 – 58(1963)).
122 See, e.g., Robert Cooter, Do Laws Make Good Citizens? An Economic Analysis of Internalizing Legal
Values, U.C. Berkeley Law and Economics, Working Paper Series, p. 16. ; Dan Kahan, What do Alternative
Sanctions Mean?, 63 U. CHI. L. REV. 591, 603 – 04 (1996); Richard McAdams, Legal Construction of Norms, 86
Toward an Economic Theory of Pro-Defendant Criminal Procedure 30
can only serve to weaken this respect because the message that the activity is
undesirable is clouded to some extent by the message that law enforcement is
selective and biased. This reduces, one would expect, whatever ability the
criminal law has to shape preferences and influence behavior outside of purely
deterrence based arguments (focussing on expected sanction arguments).123
Second, members of society may derive some utility from expressing
condemnation.124 Although this may be true in some instances the benefits
inuring from this must be weighed against the costs of rent-seeking and how
much the perception of rent-seeking reduces the utility from expressing
condemnation.125
In addition to stigma and expressive effects, rent-seeking may have
“enforcement cost” effects. If law enforcement is perceived to be biased then it is
likely that certain people will refuse to assist law enforcement. An “us” and
“them” mentality may arise making it more difficult to solicit information and
evidence.126 This would increase the difficulty and costs associated with
prosecutions, which in turn reduces the likelihood that wrongdoers would be
sanctioned. As the probability of being sanctioned decreases the deterrent effect
of the law is reduced because the expected sanction drops, all else equal.127
VI. METHODS OF CONSTRAINING RENT-SEEKING
Taken together the direct and deterrence-related costs of lobbying or rent-
seeking seem significant enough to consider methods of constraining them
subject to how much we want to spend monitoring prosecutorial behavior. In
this Part we examine two structural methods of constraining prosecutorial
behavior: procedural protections and “inefficient” punishments. We also
discuss the implications of our theory for the jury’s institutional role.
Constraining prosecutorial behavior to reduce rent-seeking related costs
is, in a sense, analogous to the efforts of corporate law in constraining the agency
VA. L. REV. 1649 (2000); Cass Sunstein, Expressive Function of Law, 144 U. PA. L. REV. 2021, 2022, 2025-29
(1996).
123 See Kahan, supra note 122; Sunstein, supra note 122.
124 See Dan M. Kahan, Punishment Incommensurability, 1 BUFF. CRIM. L.R. 691, 693 – 94; see also
Interdisciplinary Program Series, The New Chicago School Myth of Reality?, 5 U. CHI. L. SCH. ROUNDTABLE 1, 3
(1998); Paul H. Robinson & John M. Darley, The Utility of Desert, 91 Nw. U.L. REV. 453, 472 (1997).
125 See generally Kaplow & Shavell, supra note 60.
126 Cf. Nadler, supra note 116, at 32 – 41 (discussing results of certain experiments which suggest
that group identity does matter when law enforcement is perceived to be “unjust” reflected in willingness to
commit unrelated wrongs and mock juror verdicts).
127 One possibility worth noting is that if a group of people are reluctant to provide information to
law enforcement about other members in their group (absent selective enforcement issues) then that might
be a reason itself to proceed with selective enforcement in that group to increase the probability of being
sanctioned. See Kay B. Perry, Fighting Corruption At The Local Level: The Federal Government’s Reach Has Been
Broadened, 64 MO. L. REV. 157, 162 (1999).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 31
costs arising from the separation of ownership and control.128 The prosecutor
can be viewed as an agent for society in a manner similar to how a manager or
employee is often viewed as an agent for a corporation.129 However, unlike the
corporate context, environmental factors that constrain the agency costs of
private firms are not fully present in the case of governments. Since
governments do not issue stock, we do not observe discounts in their share
prices due to agency costs, nor do governments face the same risk of losing out to
competitors because of agency costs as business enterprises do.130 Hence, explicit
constraints on rent-seeking, such as pro-defendant procedural protections, play a
relatively important role in the public sector.
A. Procedural Protections
The argument for how procedural protections constrain agency costs in
the criminal process is that they make using the criminal process costly for both
prosecutors and for society. These costly protections should deter those who
would seek to transfer wealth through the criminal process relative to where the
protections were absent, and hence reduce the incentive to engage in, and the
costs associated with, rent-seeking in the criminal setting.
Procedural protections increase costs to society because they require
greater effort by the prosecutor to obtain convictions which increases the cost of
bringing prosecutions as well as potentially increasing the number of false
acquittals or increasing the number of meritorious cases not brought.131
Prosecutors, because they are often rewarded when conviction rates are high,
bear much of the brunt of the costs of these procedures because they make cases
more difficult to prosecute and win.132 The difficulty of using the criminal
128 See generally ROBERT C. CLARK, CORPORATE LAW (1986); WILLIAM A. KLEIN & J. MARK RAMSEYER,
BUSINESS ASSOCIATIONS – AGENCY, PARTNERSHIPS, AND CORPORATIONS (4th ed. 2000).
129 See Rebecca Hollander-Blumoff, Getting to “Guilty”: Plea Bargaining as Negotiation, 2 HARV.
NEGOTIATION L. REV. 115 (1997) (noting that “[a] prosecutor… is the agent of the people whom the office
purports to protect.”); Caroline Heck Miller, Knowing the Danger from the Dance: When the Prosecutor is
Punished for the Government’s Conduct, 29 STETSON L. REV. 69, 78 (1999) (stating that “[p]rosecutors are agents
of the sovereign that employs them.”).
130 See CLARK, supra note 128, Ch. 4 (1986); Tamar Frankel, Fiduciary Law, 71 CALIF. L. REV. 795, 811
(1983) (arguing that different compensation schemes and techniques should have the effect of reducing
conflicts of interest with fiduciaries.)
131 See Kahan, supra note 115, at 389 (noting that “the ‘exclusionary rule’, the Miranda doctrine, and
like constraints reduce the price of crime by lowering the probability that offenders will be convicted. The
cost of such rights, then, consists of the resulting increase in crime, or the resources that must be invested-
primarily in increased severity of punishment- to offset the rights discount on price of crime”). See generally.
Atkins & Rubin, supra note 1; Cassell, Guilty & Innocent, supra note 1; Cassell, Social Costs, supra note 1;
Donohue, supra note 1.
132 See J. Mark Ramseyer & Eric B. Rasmussen, Why if the Japanese Conviction Rate So High, 30 J.
LEGAL STUD. 53 (2001) (arguing that because of the stigma in Japan of acquitting defendants and the
corresponding detrimental career effects it may have for prosecutors and judges, they prosecute only strong
cases); Fred C. Zacharias, Structuring the Ethics of Prosecutorial Trial Practice: Can Prosecutors Do Justice, 44
Toward an Economic Theory of Pro-Defendant Criminal Procedure 32
process and the attendant costs involved makes it less useful as a wealth-
extraction tool.
Thus, a key function of pro-defendant procedural protections is to
increase the probability that a prosecutor and a bribing party will be unable to
find a mutually-acceptable bribe, thus making the set of contractible bribes zero
or close to it. This works as follows. On the prosecutor’s side, procedural
protections raise the cost of targeting innocent parties. If the prosecutor targets
innocent group B members, he is unlikely to be successful given all the pro-
defendant procedures relative to where these protections are absent.133 If he
maintains his promise to target group B, he will have very few successful
prosecutions, and will probably lose his job (as high conviction rates are
important to prosecutors). This suggests that the prosecutor will demand a very
high bribe in order to adopt a selective enforcement policy. Moreover, given the
risk of losing his job, potential bribers should doubt the credibility of the
prosecutor’s promise to selectively enforce. Given the difficulty of implementing
a successful selective enforcement policy, the doubtful credibility of the
prosecutor, and the negligible benefits to both parties, the potential briber’s
willingness-to-pay should fall substantially. 134
Procedural protections also make it more difficult for corruption to
flourish. A prosecutor who threatens to arrest individuals on false charges
would find it considerably more difficult to mount a credible threat against his
victims in the presence of pro-defendant procedural protections relative to where
these protections were absent. Hence, the prosecutor’s power to “shake down”
VAND. L. REV. 45, n. 264 (arguing that “[t]o the extent a prosecutor's conviction rate is all that counts, the
institutional incentives point toward minimizing the responsibility to ‘do justice’.”) See also Tracey L.
Meares, Rewards for Good Behavior: Influencing Prosecutorial Discretion and Conduct with Financial Incentives, 64
FORDHAM L. REV 851 (proposing a system of incentives to effect prosecutors’ conduct).
The difficulty of using the process and the attendant cost involved makes it less useful as a wealth-
extraction tool than perhaps other options and should shift lobbying and rent-seeking away, to some extent,
from the criminal process to other methods of influencing government behavior. See Paul H. Rubin,
Christopher Curran, & John Curran, Litigation Versus Legislation: Forum Shopping by Rent-Seekers, Draft 1999
(on file with authors).
Note that the use of conviction rates, as opposed to some other measure(s), for assessing
prosecutorial performance is an issue worthy of further exploration, but is currently outside the scope of this
paper.
133 But see Stuntz, supra note 14, at 27 – 28 (noting that while the constitution defines what the
criminal law process looks like, it is prosecutors and defenders who define what issues and contests to
bring) . We address some of these points and issues related to legislative lobbying in text accompanying
notes 157 – 160.
134 See A. Mitchell Polinsky & Steven Shavell, Corruption and Optimal Law Enforcement, J. PUB. ECON.
(forthcoming 2001); A. Mitchell Polinsky and Steven Shavell, On the Disutility and Discounting of Imprisonment
and the Theory of Deterrence, 28 J. LEGAL STUD. 1 (1999). See also Kaplow & Shavell, supra note 60, at 961 n. 637.
On the social costs of corruption, see ROBERT KLITGAARD, CONTROLLING CORRUPTION (1988); Susan Rose-
Ackerman, CORRUPTION: A STUDY IN POLITICAL ECONOMY (1978); Pranab Bardhan, Corruption and
Development: A Review of Issues, 35 J. ECON. LITERATURE 1320 (1997); and Andrei Schleifer & Robert W.
Vishny, Corruption, 108 Q.J. ECON. 599 (1993).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 33
individuals for money in exchange for a promise not to bring charges should be
considerably less in the regime with procedural protections. Even the
prosecutor’s ability to credibly promise not to enforce the law against a
particular defendant should fall. For if the prosecutor charges the wrong person
or no one at all, he most likely will be unsuccessful in obtaining a conviction for
this crime (given pro-defendant procedural protections). Given his difficulty in
charging and convicting an alternate candidate, the cost to the prosecutor of
promising not to enforce against a particular defendant is relatively high (i.e.,
one less conviction), and the promise probably cannot be considered completely
credible. These factors suggest the prosecutor will demand a large bribe. From
the perspective of the potential defendant (who is in the process of a shake
down), his willingness to pay a bribe falls since he is less likely to be convicted in
the first place, and any promise by the prosecutor not to enforce cannot be
regarded as completely credible.135 We can then examine some protections to
consider how they might help to constrain agency costs.
In this paper we have identified two major types of procedural protection:
the reasonable doubt standard and the double jeopardy rule. Both reduce the
prosecutor’s power to selectively enforce and hence help to constrain agency
costs. In this sense, they clearly fall within the analysis of this section because
they simultaneously raise the cost to the prosecutor of implementing a selective
policy and lower the value to the potential defendant of seeking such a policy.
The reasonable-doubt rule accomplishes this task by directly reducing the
probability of a guilty verdict and increasing the amount of evidence necessary
for conviction. The double jeopardy rule aids in this task by preventing the
prosecutor from bringing successive prosecutions against the same defendant,
with the hope of eventually learning how to convict the defendant on weak
evidence. For example, a prosecutor who loses his first case against a particular
defendant could discover that his loss was due in large measure to the weak
testimony of a non-credible witness. 136 Upon learning this, the prosecutor might
have an incentive, in the absence of a double jeopardy rule, to coach his
135 We do not discuss what might happen if the defendant suffered a large stigma simply from
being charged or indicted for certain kinds of wrongdoing. See In re Fried, 161 F.2d 453, 458 (2nd Cir. 1947)
(noting that “[f]or a wrongful indictment is no laughing matter; often it works a grievous, irreparable injury
to the person indicted. The stigma cannot be easily erased. In the public mind, the blot on a man's
escutcheon, resulting from such a public accusation of wrongdoing, is seldom wiped out by a subsequent
judgment of not guilty. Frequently, the public remembers the accusation, and still suspects guilt, even after
an acquittal.”); Eric B. Rasmusen Stigma and Self-fulfilling Expectation of Criminality, 39 J.L. & ECON. 519
(1996). In such cases the potential for corruption and wealth extraction are greater because the prosecutor
can gain or impose costs without actually having to win at a trial (i.e., without having to obtain a
conviction). These instances are simply outside the scope of this paper.
136 See Developments in the Law—Corporate Crime: Regulating Corporate Behavior Through Criminal
Sanctions, 92 HARV. L. REV. 1341-49 (1979).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 34
originally non-credible witness in order to boost his credibility before a jury in a
later trial.137
Our theory suggests that the reasonable doubt and double jeopardy rules
work hand-in-hand as complementary rules. Both rules impose important
restraints on a prosecutor who seeks to implement a selective or predatory
enforcement policy. The reasonable doubt rule reduces the probability of success
in each case of targeted enforcement. But no matter how low the probability of
success is reduced, the prosecutor may still have an incentive to adopt a selective
enforcement policy if he can bring successive actions against a particular
defendant. Suppose, for example, that the probability of conviction is the same
in every trial. In the extreme case in which the prosecutor can bring an infinite
number of successive actions against the defendant, he is very likely to
eventually get a conviction, no matter how small the probability of conviction in
the individual trial.138 The more worrisome case, however, is where the
prosecutor learns from a previous mistake and uses the information from a “test
trial” to boost the probability of conviction to a near certainty in the second
trial.139 In either of these cases (fixed probability of conviction or increasing
probability of conviction), the reasonable-doubt rule alone is probably
insufficient to dampen the prosecutor’s incentive to selectively enforce. The
reasonable doubt rule should be coupled with a double-jeopardy-like rule to
substantially dampen the prosecutor’s incentive to selectively enforce the law.140
B. Penalty Restrictions or “Inefficient Punishments”
Another way to constrain the costs associated with abuses of prosecutorial
or governmental authority is to utilize restrictions on the size of penalties or the
process by which they are levied. David Friedman has described the size
restrictions as “inefficient” punishments.141 This epithet is not meant to be
pejorative. Friedman distinguishes “inefficient” punishments, like prison, from
“efficient” punishments, like the death penalty administered quickly or a large
monetary penalty equal to the defendant’s wealth.142 The argument is that
“efficient” punishments may not impose large direct costs on the state and hence
137 See id.
138 Suppose the probability of conviction in one trial is p. If the prosecutor can bring an infinite
number of successive actions, each with the same probability of conviction, the likelihood of eventual
conviction is p + (1-p)p + (1-p)2p + … + (1-p)Np, which approaches 1 as N approaches infinity.
139 See Developments, supra note 136.
140 This statement is not meant to exclude alternate means of perhaps restraining prosecutorial
retrials. For example, if in the initial trial the reasonable doubt standard required 95% certainty of guilt
before conviction then we could require a 96% certainty of guilt in retrial number 1. If there was a second
retrial then we could require a 97% likelihood of guilt and so forth. Even then, given the argument in supra
note 138 it is doubtful that much would be gained through this approach.
141 See Friedman, supra note 18, at 259.
142 See id., at 260 – 61.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 35
prosecutors may be willing to use them to extract wealth from defendants.143
However, “inefficient” punishments impose greater costs on the state than
“efficient” punishments (e.g., the costs of maintaining prisons) and hence reduce
the incentive of prosecutors and the state to use the criminal process to extract
wealth.144 Simply put, “inefficient” punishments increase the costs of using the
criminal process relative to “efficient” punishments for the state thereby
reducing the incentive of the prosecutor to use the criminal process for self-
interested ends relative to where only “efficient” punishments were available. 145
At least two types of penalty restrictions are relevant to this analysis: the
prohibition of retroactive punishments and the prohibition on cruel and unusual
punishments.146 The latter restriction fits comfortably with the analysis here as
well as Friedman’s. The state could easily adopt a low-cost (for itself) system of
punishment. Defendants could be executed, enslaved, or put into laboratories
for scientific experimentation and the harvesting of organs and tissue. 147 Instead,
we observe a system, in which the state forgoes the opportunity to extract all of
the defendant’s wealth in the form of a penalty, and prison terms force the state
to forgo the full value of the convict’s labor. The constitutional prohibition on
cruel and unusual punishments is in part responsible for this choice, although
the choice seems to have been made in some countries where there is no such
prohibition.148
Consistent with Friedman’s argument, our analysis suggests that the
adoption of “inefficient punishments” increases the cost of punishment to the
state, relative to “efficient punishments”, and at the same time reduces the
amount a potential defendant would be willing to pay in order to avoid being
charged with a crime. Since the latter effect is obvious, only the former effect
deserves some discussion. The prohibition on cruel and unusual punishments
reduces the potential incentive for the state or the prosecutor to punish innocent
individuals in order to profit from their punishment. By raising the cost of
punishment, or simply making punishment costly, the prohibition enhances the
143 See id., at 261.
144 See id., at 263 – 64.
145 See id., at 263.
146 See U.S. CONST. ART I, § 9, CL. 3 (stating “No Bill of Attainder or ex post facto Law shall be
passed.”) A similar prohibition applies to the states: “No State shall . . . pass any Bill of Attainder . . ..” Id.,
ART. I, § 10, CL. 1; U.S. CONST. AMEND. VIII (stating “Excessive bail shall not be required, nor excessive fines
imposed, nor cruel and unusual punishments inflicted.”).
147 See Friedman, supra note 18.
148 For example, Malaysia, Morocco, Senegal, and the Ivory Coast do not have a rule prohibiting
cruel and unusual punishment. Yet, the state in those countries has apparently not gone to the extreme of
trying to profit from punishing the guilty. For information on Transparency International’s Corruption
Perceptions Index, see .
Toward an Economic Theory of Pro-Defendant Criminal Procedure 36
likelihood that the state will punish only the guilty.149 It also dampens incentives
individuals might have to become prosecutors in order to enrich themselves.
The prohibition of retroactive punishments – a restriction on the penalty
imposition process – constrains rent-seeking at the legislative level. In the
absence of such a restriction, interest groups could use the criminal process to
confiscate the wealth of other groups or of particular individuals. A predatory
enforcement regime could retroactively impose a criminal penalty on the activity
of a particular group and use the new law as leverage to expropriate their
wealth. The ex post facto and bill of attainder clauses in the Constitution both
apply to this type of activity. The ex post facto clause applies specifically to
legislative attempts to punish retroactively.150 The bill of attainder clause applies
to legislative attempts to punish particular individuals without a trial.151 As we
will argue below, our approach to criminal procedure provides some insight into
the doctrines courts have developed to deal with these challenges.
Although Friedman’s analysis of inefficient punishments is generally
consistent with ours, it does raise an interesting question – why do we need
procedural protections when we could use inefficient punishments or vice versa?
For example, if we thought our current level of inefficient punishments had not
sufficiently reduced the costs associated with abuses of prosecutorial authority
we could simply make punishment even more inefficient rather than creating a
pro-defendant bias in criminal procedure. The presence of both methods of
constraining prosecutorial behavior suggests that that there may be some
advantages that the pro-defendant procedures possess that make them attractive
relative to simply making punishments more inefficient.
The key difference between inefficient punishments and procedural
protections has to do with how much of the cost of each method (the procedure
or the punishment) the prosecutor bears. The costs of inefficient punishments
appear to be largely borne by the state and society, but not particularly by the
149 The reason for why the incentive to go after the truly guilty increases because of inefficient
punishments requires some explanation. An inefficient punishment increases the costs of convicting anyone
(guilty or innocent), and hence increases the costs of convicting both types of defendants. This should result
in the state (and the prosecutor) shifting more resources towards targeting the truly guilty rather than
bringing cases against the innocent as the expected payoffs from convicting the truly guilty are probably
higher than the expected payoffs from convicting the innocent as a general matter.
150 See Harold J. Krent, The Puzzling Boundary Between Criminal and Civil Retroactive Lawmaking, 84
GEO. L.J. 2143 (1996); Wayne A. Logan, The Ex Post Facto Clause and The Jurisprudence of Punishment, 35 AM.
CRIM. L. REV. 1261 (1998).
151 See Jane Welsh, The Bill of Attainder Clause: An Unqualified Guarantee of Process, 50 BROOK. L. REV.
77 (1983); Thomas B. Griffin, Beyond Process: A Substantive Rationale for the Bill of Attainder Clause, 70 VA. L.
REV. 475 (1984). We will focus our discussion infra on the ex post facto clause for the sake of brevity. Note
that both clauses induce a reactive type of rent seeking, as they seem to invite defendants to challenge
virtually every effort to punish on the ground that it is either a disguised bill of attainder or retroactive
penalty.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 37
prosecution. They may impose only an indirect cost on the prosecution.
However, pro-defendant procedures impose more direct costs on prosecutors
because they make obtaining convictions more difficult (i.e., the probability of
winning decreases). Given that high conviction rates are important to many
prosecutors for career reasons, and perhaps personal ones, we would expect
prosecutors to respond more to these more direct measures than to indirect
concerns about how much a punishment is costing the state.152 This is especially
so when many prosecutors are paid by local authorities and prisons are funded
by state authorities.153 Consequently, one might view procedural protections as
being more direct means of influencing prosecutorial behavior than simply
making punishments inefficient.154
This distinction between inefficient punishments (or more generally,
penalty restrictions) and procedural protections suggests another sense in which
rules designed to dampen rent-seeking may serve as complements. The
procedural protections described earlier – the reasonable-doubt rule and the
double-jeopardy rule – may work well in the short run in removing the
incentives for the prosecutor to selectively enforce the law. However, if the
background institutional structure is one that allows the state to profit in some
sense from the punishment of individuals, we should worry about how long the
prosecutor will be able to stay out of the predatory enforcement game. Just as
the potential for profit induces entry of new businesses in the private sector, the
potential for profit in enforcement should induce entry of a similar sort in the
public sector. Creative prosecutors would find ways to modify the procedural
rules, plea bargain around them, or to lobby the legislature until the desired
changes were enacted. Eventually the procedural protections would be watered-
down to a point that would enable self-interested enforcement agents, and their
support coalitions, to reap the rewards from selective enforcement.155 Thus,
inefficient punishments and procedural protections can work as complements. 156
152 See Sanford I. Weisburst, Judicial Review of Settlements and Consent Decrees: An Economic Analysis,
28 J. LEGAL STUD. 55, 88 (1999) (arguing that if a prosecutor feels his efforts will escape public notice, the
prosecutor will expend less personal effort and the result will not be aligned with the public’s desire to
punish). In some respects one can view the procedural protections as alignment measures – trying to align
prosecutors’ interests with social welfare (or the benevolent state), whereas the inefficient punishments
might be seen as constraints on a non-benevolent state.
153 See Stuntz, supra note 10, at 20 – 37 (discussing the incentives of the different actors in law
enforcement).
154 Note that if punishments were “inefficient” then the state may want to constrain prosecutorial
behavior and may do so by using pro-defendant procedural protections. However, it is not necessary for
the state to have “inefficient” punishments before it might consider relying on procedural protections.
155 See Stuntz, supra note 14, at 26 (arguing that despite pro-defendant protections, prosecutors have
held conviction rates constant and lowered the average cost of prosecution by prosecuting “winning” cases).
See Stuntz, supra note 10, at 6 – 15 (arguing that because criminal codes are so broad it gives a prosecutor the
ability to selectively enforce).
156 One point to note is that procedural protections, inefficient punishments, and other measures
may work as substitutes to some extent as well. There may be some overlap with each kind of measure
(even though there are differences too). Which is the optimal balance of measures is outside the scope of
Toward an Economic Theory of Pro-Defendant Criminal Procedure 38
This also suggests why we would not want to rely exclusively on procedural
protections to reduce rent-seeking.
One might also ask why we should bother to have procedural protections
if all that might happen is prosecutors find ways around them or groups in
society substitute increased legislative lobbying for the now more costly
prosecutorial lobbying. 157 Our response is two-fold. First, we do not suggest
that procedural protections will perfectly align prosecutor’s interests with social
welfare. We are simply claiming that pro-defendant procedural protections
reduce the incidence and costs of rent-seeking in the criminal enforcement
process relative to where these protections are absent.158 Second, an increase in
legislative lobbying due to procedural protections does not mean that there has
been no improvement in social welfare as a result of the procedural protections.
Absent any kinds of constraints we would expect interest groups to lobby at all
levels of the government process (e.g., investigation and enforcement,
adjudication, legislation) until the marginal benefits from each kind of lobbying
this current paper, but is a matter worthy of greater inquiry. Such matters invoke questions about things
like why we should have these 3 procedures and this magnitude of inefficiency in punishments as opposed
to, say, 4 procedures and a different magnitude of inefficiency in punishments.
In addition to the procedures discussed above there may be other areas of law that have some
tendency to curtail self-interested behavior by prosecutors even though that may not be the primary reason
for that area of law. Consider, for example, intent or mens rea requirements in the criminal law. See Cass &
Hylton, supra note 18, at 45 – 53 (suggesting that intent requirements may constrain rent-seeking behavior as
they limit the ability of litigants (whether plaintiffs or prosecutors) to threaten to bring suits in the antitrust
context). By making conviction harder to obtain (due to the requirement of proving mens rea) we reduce
the incentive of prosecutors to use the criminal law to benefit themselves. See Mens Rea In Federal Criminal
Law 111 HARV. L. REV. 2402, 2419 (1998). Although this may sometimes be a benefit of intent requirements
we argue that these requirements are different to the procedural protections we are considering in a number
of ways.
First, intent standards are subject to some change by the legislature (witness the corporate crime
area which has many strict liability offenses and over 120 mens rea standards at the federal level alone),
whereas the Constitutional procedural protections are not that easy to change by the legislature. See William
A. Laufer, Culpability and the Sentencing of Corporations, 71 NEB. L. REV. 1049 (1992). Second, even if the
intent requirement is not met there may be a lesser-included offense of which the defendant might be guilty.
See Theodore E. Lauer, Burglary In Wyoming, 32 LAND & WATER L. REV. 721, 786 (1997). If so, the prosecutor
can still threaten the defendant with that lesser offense (although since it probably carries a less severe
penalty than the mens rea offense the threat may not be as strong). See Stuntz, supra note 14, at 7 (noting that
the use of simple traffic laws to procure reasonable suspicion and sodomy laws in sex cases are means that
prosecutors use to punish defendants without proving the real crime). However, if there is a violation of a
procedural protection it will be difficult to convict the defendant at all (absent harmless error or some other
exception). See DRESSLER, supra note 4, at 57 – 61 (discussing harmless error doctrine). Third, intent
requirements may be justified by many other things besides a concern with rent-seeking, whereas
procedural protections appear more clearly designed to target abusive behavior by prosecutors. See Jeffrey
S. Parker, The Economics of Mens Rea, 79 VA. L. REV. 741 passim (1993); Steven Shavell, Criminal Law and the
Optimal Use of Nonmonetary Sanctions as a Deterrent, 85 COLUM. L. REV. 1232, 1247 – 49 (1985). Thus, although
there may be other ways to reduce rent-seeking costs it would appear that pro-defendant procedural
protections may be more direct and effective means of constraining these costs than exclusive reliance on
other means of impacting prosecutorial behavior.
157 See Stuntz, supra note 14, at 27 – 28, 30.
158 See id.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 39
equaled the marginal costs.159 This division of lobbying efforts should provide
the highest returns to interest groups. Further, it seems safe to assume that
lobbying is subject to diminishing returns (i.e., after a certain point the payoffs
from lobbying in one sphere begin to diminish relative to earlier expenditures).
When we impose pro-defendant constraints we make the costs of rent-seeking
higher in the criminal enforcement context and drive some of the lobbying in that
sphere into some other activity (e.g., lobbying for legislation). By assumption,
this reduces the gains to the interest groups relative to where there were no
protections.160
C. Some Implications for the Jury
Although we have focused on rent-seeking in the enforcement process
rather than in the legislative process, the jury serves as an important constraint
against both types of rent-seeking.161 A prosecutor who brings politically
motivated charges against members of politically weak groups faces the risk,
under the jury system, of being unable to gain a unanimous verdict from a jury
consisting of some members from the weak group. Indeed, the theory of this
paper suggests an important rationale for the requirement of unanimity among
jurors in criminal trials. The need to obtain a unanimous verdict makes it more
difficult for the prosecutor to selectively target politically marginal groups or
159 See Khanna, supra note 35, Part V (discussing a similar argument in the context of expenditures
at different trial and appeal levels).
160 Assume that if there were no constraints interest groups would expend resources in the amount
of $100 in legislative lobbying and $100 in enforcement lobbying resulting in a gain of $250 to the groups (a
net gain of $50). When we impose pro-defendant constraints what we do is to make the costs of rent-seeking
higher in the criminal enforcement context and drive some of the lobbying in that sphere into some other
activity (e.g., lobbying for legislation). By assumption, this reduces the gains to the interest groups (the pre-
constraint equilibrium produced the highest net gain to the groups) relative to where there were no
protections. Thus, now we may have $150 spent on legislative lobbying and $50 in enforcement lobbying
providing a gain of $205. In other words, the abuse defendants suffer (i.e., the wealth extracted) in total
should be less than without the protections (i.e., $205 versus $225). Whether this amount should be reduced
further, how, and at what cost is outside the scope of this paper. See Stuntz, supra note 14, at 75 – 76
(discussing how the constitution should be restructured to prevent abuses to criminal defendants). See also
William J. Stuntz, Self-Defeating Crimes, 86 VA. L. REV. 1871, 1891-95 (2000) (discussing how prosecutorial
discretion undermines the goals of criminalization). Our point is that just because procedural protections
force a shift in lobbying activity to a different level of government (i.e., from enforcement to legislation) does
not mean that the harm to society is the same – it is still less than when the constraints were absent. Further,
some of the procedural protections and constitutional law doctrines are targeted at making rent-seeking in
the legislative process more difficult (e.g., void-for-vagueness and penalty restrictions). Penalty restrictions
work to impair rent-seeking because “…the cost that enforcers can impose on defendants is less and because
the cost to the enforcement system of carrying out the threat is greater.” Friedman, supra note 18, at 268.
161 See Duncan v. Louisiana, 391 U.S. 145, 156 (1968) (arguing that “[p]roviding an accused with the
right to be tried by a jury of his peers gave him an inestimable safeguard against the corrupt or overzealous
prosecutor and against the compliant, biased, or eccentric judge.”); Bennett L. Gershman, The New
Prosecutors, 53 U. PITT. L. REV. 393, n. 14 (1992) (arguing that because the new breed of prosecutors is less
accountable to society and his peers, the jury may stand as the as the only effective check on the
prosecutorial power.)
Toward an Economic Theory of Pro-Defendant Criminal Procedure 40
individuals in the law enforcement process. 162 The need to obtain a unanimous
verdict from the jury also gives the jury the power to nullify statutes designed to
expropriate wealth from politically marginal groups.
Our theory also provides some insight into the original function of
challenges to the jury’s composition, including the controversial problem of
peremptory challenges. Challenges may be put to either the whole array of
jurors or to individual jurors. Blackstone explains that
challenges to the array are at once an exception to the whole panel,
… and they may be made upon account of partiality or some
default in the sheriff, or his under-officer who arrayed the panel. . .
Also, though there be no personal objection against the sheriff, yet
if he arrays the panel at the nomination, or under the direction of
either party, this is good cause of challenge to the array.163
In other words, the fundamental common law rationale for permitting challenges
to the whole jury is the suspicion, no doubt grounded on evidence, that the
sheriff chose the jurors in order to maximize his chances of obtaining a
conviction. Challenges to the whole array were apparently permitted to prevent
the sheriff from implementing a selective enforcement policy.
Challenges to individual jurors could be based on cause, or could be
peremptory, in the sense of not being based on any of the accepted grounds. 164
Peremptory challenges were granted only to the defendant.165 Although
peremptory challenges have come under attack more recently as a form of
invidious discrimination, the original purpose is somewhat easier to see in the
context of a rent-seeking model.166 One could view the peremptory, in this
analysis, as giving the defendant a zone of unquestioned authority in the choice
of jurors, so long as he did not use it to an excessive degree. If a wily predatory
sheriff had managed to choose conviction-prone jurors in a way that would be
difficult to challenge on the accepted grounds, the defendant could always fall
back on his peremptory challenges. To the extent that this obstruction stood in
162 An alternative explanation for the unanimity requirement – that it reduces false convictions may
not be terribly convincing. See Feddersen & Pesendorfer, supra note 64 (arguing that under plausible
assumptions the unanimity requirement may result in an increase in false convictions relative to a
supermajority vote requirement).
163 See BLACKSTONE, COMMENTARIES, supra note 24, at 359.
164 See id., at 361 – 63.
165 See id., at 362.
166 See J.E.B. v. Alabama ex rel. T.B., 511 U.S. 127, 140 (1994) (holding that the Equal Protection
Clause prohibits gender discrimination in the use of peremptory challenges); Batson v. Kentucky, 476 U.S.
79, 89 (1986) (holding the prosecutor’s peremptory challenges based solely on race were unconstitutional
under the Equal Protection Clause.)
Toward an Economic Theory of Pro-Defendant Criminal Procedure 41
the way of any effort to selectively enforce the law, the sheriff would have a
much smaller incentive to try to control the composition of the jury.167
If, as we have argued, constraining rent-seeking costs provides the core
justification for procedural protections then it becomes important to consider
how this justification squares with some details of legal doctrine. We consider
this in the next Part.
167 In addition, when the right to a jury trial is available may be consistent with a rent-seeking
approach as well. Jury trials are available as of right for most criminal cases except those that carry trivial or
fairly small penalties. See Duncan v. Louisiana, 391 U.S. 145, 160 (1968) (holding that “there is a category of
petty crimes or offenses which is not subject to the 6th Amendment jury trial provisions”). This is consistent
with rent-seeking because trials that carry very small or trivial penalties may not be particularly attractive
means with which to extract wealth for prosecutors. The threat the prosecutor can generate with paltry
penalties is quite small and hence so is the concern with rent-seeking relative to where penalties are larger.
See Friedman, supra note 18, at 268. For such small sanctions the costs of the jury trial are probably not
justified by any reduction in rent-seeking (which is probably small in this context).
Further, alternative explanations for the right to a jury trial do not appear to provide as complete a
picture as they might if they considered concerns with rent-seeking. One potential explanation for the right
to a jury trial is that society values the expression of the popular will as reflected in a jury decision. See
Spaziano v. Florida, 468 U.S. 447, (1984) (Stevens, J. dissenting in part) (arguing that the right to have an
authentic representation of the community’s views on the determination that must precede a deprivation of
liberty supports the constitutional entitlement to a trial by jury). If this were the only purpose behind the
right to a jury trial we would expect all jury decisions to be unreviewable. However, this is not the case
because the law permits jury convictions to be appealed but not jury acquittals. See Kepner v. United States.,
195 U.S. 100 (1903). See also Steinglass, supra note 35, at 354 – 55 (1998).
Another potential explanation for the right to a jury trial is that it either reduces erroneous
decisions relative to bench trials or is less likely to falsely convict relative to bench trials. See HARRY KALVEN,
JR. & HANS ZEISEL, THE AMERICAN JURY 6-7 (1971). This argument is not particularly convincing because it is
a little difficult to believe that jury trials are likely to be more accurate (i.e., less error prone) than bench
trials. One doubts there is any empirical evidence to support this result and our legal system also seems to
suggest that jury trials may be more prone to errors than bench trials. See VALERIE P. HANS & NEIL VIDMAR,
JUDGING THE JURY 126-27 (1986) (stating that “[w]e are thus led to the conclusion that jurors may not always
be able to follow the law as it is intended to be.”). Much of the law of evidence seems to try to protect the
jury from misperceptions and bias, whereas we seem less concerned with these matters for bench trials. See,
e.g., FED. R. EVID. 103(C), 403. This suggests bench trials are probably more accurate than jury trials overall
or at least not less accurate as a general matter.
It may be, however, that we believe juries are less likely to falsely convict compared to bench trials.
It is not entirely clear why we would believe this if we think bench trials are generally more accurate.
Perhaps the argument is that judges are more biased against defendants than a jury of the defendant’s peers
as judges tend to be in quite a different socio—economic strata as compared to most defendants. See HANS &
VIDMAR, supra (noting that “[f]or criminal trials the pattern disagreement shows that the jury was usually
more lenient toward the defendant than was the judge.”); Pnina Lahav, The Chicago Conspiracy Trial:
Character and Judicial Discretion, 71 U. COLO. L. REV. 1327, 1340 (2000). There may be some empirical evidence
supporting the differing rates of false convictions (or maybe tied to it). See generally KALVEN & ZEISEL, supra.
Perhaps this is true, but when phrased this way it appears more consistent with concerns about rent-
seeking. This is because this suggests that judges as a group may discriminate (i.e., use the criminal process)
against criminal defendants as a group. If so, then this justification squares well with a rent-seeking
approach. Also even if jury trials result in fewer false convictions and more false acquittals than bench trials
the issue is raised about whether the asymmetry is desirable. It may be too severe on traditional error cost
grounds for the same reason that the reasonable doubt standard may be too severe on traditional error cost
grounds.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 42
VII. APPLICATIONS OF POSITIVE THEORY
We have focused on three major types of procedural protections: the
reasonable doubt standard, the double jeopardy rule, and the right to a jury trial.
We have also discussed “penalty restrictions”, such as rules against cruel and
unusual punishment, ex post facto punishment, and bills of attainder. However,
we have been concerned so far with explaining broad institutional features. In
this Part we extend the argument by taking a more detailed look at the case law
associated with some of these pro-defendant protections. Since criminal
procedure is a vast area of the law, we will provide only a sketchy analysis here.
We claim that the theory developed in this paper provides a good positive theory
of criminal procedure doctrine.
A. Double Jeopardy
As a general matter, Double Jeopardy reduces the prosecutor’s power to
selectively enforce or abuse his discretion in a manner complementary to the
reasonable doubt standard. Double Jeopardy complements the reasonable doubt
rule by preventing the prosecutor from bringing successive prosecutions against
the same defendant, with the hope of eventually learning how to convict the
defendant on weak evidence. 168 Not only does the overall structure of Double
Jeopardy seem to reflect rent-seeking concerns, but also some of the different
aspects of Double Jeopardy seem to reflect this concern as well.169 We focus on
three aspects to highlight how Double Jeopardy displays concerns with
controlling abuse of the criminal process by prosecutors in addition to being
concerned with the number and types of errors.
One aspect of Double Jeopardy that has attracted some attention is the
treatment of appeal rights. In Kepner v. U.S., the Supreme Court held that appeal
rights are asymmetric,170 in the sense that the defense generally can appeal any
conviction, but the prosecution’s right to appeal acquittals is severely limited. 171
On its face, this rule seems like it might reduce the number of false convictions
and increase the number of false acquittals relative to symmetric appeal rights
because it denies the prosecution the ability to correct false acquittals at the trial
168 See Benton v. Maryland, 395 U.S. 784, 796 (1969) (noting that the rationale for double jeopardy
includes the policy against allowing multiple prosecutions that will enhance the possibility that the innocent
may be found guilty).
169 See Stith, supra note 35.
170 See Kepner v. United States., 195 U.S. 100, 105 (1904).
171 See Sanabria v. United States, 437 U.S. 54, 64 (1978) (holding that even if legal rulings on the
exclusion of evidence leading to acquittals were erroneous the prosecution could not appeal); Fong Foo v.
United States, 369 U.S. 141, 143 (1962) (holding that the prosecution could not appeal an acquittal where the
judge, who lacked the authority to do so, directed a verdict of acquittal before the prosecutor has rested his
case); Carroll v. United States, 354 U.S. 394, 400 (1956); United States v. Ball, 163 U.S. 662 (1896) (holding that
a defendant may not be prosecuted more than once for an offense); Stith, supra note 35, at 13, 18.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 43
level, while also denying the prosecution the ability to turn correct acquittals into
false convictions through the appeals and retrial process.172 However, prior
analysis reveals that, on closer inspection, this is not necessarily the case.173 This
is because not only are the above mentioned effects possible, but also so are some
countervailing effects. For example, by giving the prosecution only one shot at
obtaining a conviction (i.e., the initial trial) we may provide the prosecution with
an incentive to increase spending in the initial trial relative to spending in the
initial trial under a symmetric appeals rights regime.174 This may lead, all else
equal, to an increase in the number of convictions (and perhaps false convictions)
in the initial trial relative to symmetric appeal rights.175 Thus, the effects of
asymmetric appeal rights on false convictions may not be very clear or
significant.176 Similar arguments suggest that false acquittal effects are likely to
be small and probably ambiguous.177 The overall result is that it is not at all clear
whether false convictions or false acquittals are actually reduced or increased
(much) as result of asymmetric appeal rights. 178
In light of this, it seems doubtful that traditional error-cost arguments are
a sufficient basis for asymmetric appeal rights. However, if we add concerns
with rent-seeking and self-interested prosecutors then a stronger rationale for
asymmetric appeal rights emerges.179 The asymmetric appeal rights rule of
Kepner has the effect of making the jury’s initial determination of acquittal final.
By denying prosecutors the option to have a jury’s acquittal determination
reviewed by an appellate court, the Kepner asymmetry rule enhances the power
of the jury relative to that of the prosecutor. 180 Given the unanimity requirement
and the jury’s composition after the defendant’s challenges, the Kepner rule
172 See Khanna, supra note 35, at 29 – 30.
173 See id., at 22 - 53.
174 See id., at 22 – 47.
175 See id.
176 Note that given that there are few acquittals (both in symmetric and asymmetric appeal rights
jurisdictions) and that where the prosecutor is allowed to appeal they do so infrequently, one suspects that
the false convictions reducing or increasing effects are likely to be small because few resources are being
saved by prohibiting the few prosecutorial appeals that might arise under symmetric appeal rights. See id.,
at 33 n. 120, 39 – 40.
177 See id., at 47 – 48. Under asymmetric appeal rights at least two effects on false acquittals are
possible. First, we might increase false acquittals because the prosecution cannot appeal erroneous
acquittals. Symmetric appeal rights would permit some of these incorrect acquittals to be appealed and
presumably corrected. Second, we might decrease false acquittals because if the prosecution does spend
more in the initial trial under asymmetric appeal rights then acquittals obtained in these hard-fought initial
trials are more likely to be correct ones, relative to acquittals obtained under symmetric appeal rights
regimes. Although the net effects depends on many factors, the crucial point is that false acquittals might
not unambiguously increase. See id.
178 Thus, it may be difficult to say, as a general matter, which effect will dominate. See id., at 52 – 53.
179 See id., at 68 – 70.
180 Cf. Westen & Drubel, supra note 35, at 122 – 55 (discussing the role of jury nullification in the
context of asymmetric appeal rights).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 44
increases the difficulty facing any prosecutor who mounts a selective
enforcement campaign.
Justice Holmes’s dissent in Kepner is instructive largely because it focused
on the wrong theory for asymmetric appeal rights. Holmes argued that the
majority’s decision in Kepner made no sense if understood as a rule preventing
retrials, because some retrials would occur following a successful defense appeal
of a conviction. 181 However, this is an incomplete and inadequate rationale for
the decision in Kepner. The theory we advance provides a superior rationale for
the outcome in Kepner. Under our theory the Double Jeopardy rule is not
designed for the sole purpose of controlling or preventing retrials. Its purpose is
to prevent prosecutors from successfully implementing a selective or targeted
enforcement policy. A regime in which prosecutors could appeal acquittals ad
infinitum would be much more vulnerable to selective enforcement pressures
than one where they could not.182
Another area of Double Jeopardy that appears to evince some concern
with prosecutorial abuse is the treatment of mistrials.183 The kind of abuse we
are concerned with here is that the prosecution may think, at some point in the
initial trial, that a conviction is not very likely and may then try to have a mistrial
declared by the court to try to get another shot at the defendant.184 If we
permitted the prosecution to do this and bring another trial then the prosecution
would have tremendous potential to abuse the process by having mistrials
declared whenever the prosecution thought it might not win the initial trial. This
may increase the incentive to engage in selective enforcement and induce
considerable rent-seeking behavior.185
The law appears to reflect these concerns in the way in which it addresses
whether another trial will be permitted following a mistrial. One could
characterize the law’s approach to this problem as one that depends greatly on
the defense’s attitude towards a mistrial. Thus, if the defense seeks, or does not
oppose a motion for, a mistrial then the prosecution will normally be permitted
to bring another suit. 186 This is consistent with our approach because if the
181 See Kepner v. United States, 195 U.S. 100, 134-35 (Holmes, J., dissenting) (1903).
182 See Lockhart v. Nelsen, 488 U.S. 33, 42 (1988); Ashe v. Swenson, 397 U.S. 82, 91 (1978); United
States v. Scott, 437 U.S. 82, 91 (1978). This is similar to the argument in supra note 138.
183 See Steinglass, supra note 35, at 359.
184 See Ashe v. Swenson, 397 U.S. 436, 447 (1970); Stephen Schulhofer, Jeopardy and Mistrials, 125 U.
PA. L. REV. 449, 468-69 (1977).
185 This concern is very similar to that discussed in the context of asymmetric appeal rights. At the
same time if we never permitted the prosecution to bring another trial after a mistrial we would give the
defendant a great deal of strategic power to inject matters that might lead to a mistrial when a conviction
appears likely. See generally Vikramaditya S. Khanna, The Mystery of Mistrials (Draft 2001).
186 See United States v. DiFrancesco, 449 U.S. 117, 130 (1980); see also Oregon v. Kennedy, 456 U.S.
667, 676 (1982).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 45
defense is seeking a mistrial the chances are that the prosecution is not likely to be
using the mistrial process to seek another trial to go after the defendant relative to
where the prosecution initiates the mistrial over defense objections.187 An
exception to this is where the defense seeks a mistrial based on something the
prosecution did that appears deliberately calculated by the prosecution to induce
the defense to seek a mistrial.188 This is also consistent with our rent-seeking
approach because in such cases the prospect for abuse in a second trial is fairly
high relative to where the prosecutor did not induce the defense’s motion for a
mistrial.189
On the other hand, when the defense opposes a mistrial motion the courts
have adopted a more cautious stance to permitting another trial – the
prosecution must prove a “manifest necessity” for the next trial.190 This is
generally consistent with our approach because the prospects for prosecutorial
187 See Schulhofer, supra note 184, at 468-69.
188 See Arizona v. Washington, 434 U.S. 497, 508 (1978); See also Akhil Reed Amar & Jonathan L.
Marcus, Double Jeopardy Law After Rodney King, 95 COLUM. L. REV. 1, 53 (1995). In Oregon v. Kennedy, the
Supreme Court held that the prosecutor must intend to provoke a mistrial motion in order for the defendant
to be free from reprosecution. See Kennedy, supra note 186, at 679. In Kennedy, the prosecutor asked a witness
whether he did not do business with the defendant because he was a crook and the Court found this to be
bad faith conduct on the part of the prosecution, but not enough to invoke the reprosecution exception. See
id. See United States v. Dinitz, 424 U.S. 600, 611 (1976) (stating that “[the Double Jeopardy law] bar retrials
where ‘bad-faith’ conduct by judge or prosecutor threatens the harassment of an accused by successive
prosecutions [or] a more favorable opportunity to convict the defendant;[W]here a defendant's mistrial
motion is necessitated by judicial or prosecutorial impropriety designed to avoid an acquittal, reprosecution
might well be barred”); United States v. Jorn, 400 U.S. 470, 485 (1971) (stating that “the defendant has a
significant interest in the decision whether or not to take the case from the jury when circumstances occur
which might be thought to warrant a declaration of mistrial. Thus, where circumstances develop not
attributable to prosecutorial or judicial overreaching, a motion by the defendant for mistrial is ordinarily
assumed to remove any barrier to reprosecution, even if the defendant's motion is necessitated by
prosecutorial or judicial error”).
189 See Commonwealth v. Starks, 416 A.2d 498, 500 (1980) (arguing that prosecutorial overreaching
“… signals the breakdown of the integrity of the judicial proceeding, and represents the type of
prosecutorial tactic which the double jeopardy clause was designed to protect against.”).
This exception to the bar on the application of double jeopardy where a defendant seeks a mistrial
is a very narrow one. See Green v. United States, 451 U.S. 929, 931 (1981)(Marshall, J., dissenting) (stating
that “I suspect that a defendant seeking to prevent a retrial will seldom be able to prove the Government’s
actual motivation.”). Few courts have found intentional prosecutorial inducement. In Commonwealth v.
Warfield, the Supreme Court of Pennsylvania found the requisite proof of prosecutorial intent in order to
justify an application of double jeopardy. See Commonwealth v. Warfield, 227 A.2d 177 (Pa. 1967).
Defendant was indicted for murder and voluntary manslaughter. See id., at 178. The trial judge suppressed
the defendant’s confession on constitutional grounds. See id., at 178. However, the District Attorney revealed
in his opening statement that the defendant had made a confession to the police. See id., at 178.
Subsequently, the defendant moved for a mistrial. See id., at 179. The District Attorney admitted that he
sought to induce the defendant to seek a mistrial so that the Supreme Court of Pennsylvania would rule on
the suppression of the confession. See id., at 178 – 79. The court found that, under the double jeopardy
clause, the defendant could not be tried again for first-degree murder. See id., at 180 – 81. However, the
court’s ruling did not preclude the Commonwealth from trying the defendant for second-degree murder or
voluntary manslaughter. See id., at 181.
190 See Steinglass, supra note 35, at 360. See also Arizona v. Washington, supra note 188, at 505;
WAYNE R. LAFAVE, ET AL., CRIMINAL PROCEDURE 1176-80 (3d ed. 2000).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 46
abuse of the mistrial process are greater when the prosecution seeks a mistrial
over defense objections, as compared to when the defense seeks or supports a
mistrial.191 In addition, the factors that go to showing whether a “manifest
necessity” is present largely appear to ascertain whether the prosecutor was
trying to abuse the criminal process or get a mistrial in order to avoid a loss in
the initial trial. 192 For example, a hung jury leading to a mistrial does not
present the same specter of potential prosecutorial abuse as might the injection of
prejudicial error by the prosecutor to obtain a mistrial.193 In the former case the
prosecution is often granted another trial while the latter case will normally not
result in another trial. 194 Further, when the reason for the mistrial was a move by
the defense, without prosecutorial provocation, then the scope for prosecutorial
abuse is also low and another trial is usually granted.195 These factors are all
consistent with a constraining rent-seeking approach.196
Finally, consider the on-going debate about whether Double Jeopardy
protections should apply to nominally “civil” suits brought by government
agencies that otherwise appear “punitive”.197 The courts have generally not
permitted Double Jeopardy protections to apply to nominally “civil” suits,
however, if it can be shown that the “civil” suit is in reality a form of
“punishment” then Double Jeopardy protections may apply.198 The courts seem
to rely in some measure on the following factors to determine if a “civil” suit is in
reality “punishment”:
(1) whether the sanction involves an affirmative disability or
restraint; (2) whether it has historically been regarded as a
punishment; (3) whether it comes into play only on a finding of
scienter; (4) whether its operation will promote the traditional aims
of punishment-retribution and deterrence; (5) whether the behavior
to which it applies is already a crime; (6) whether an alternative
purpose to which it may rationally be connected is assignable to it;
191 See Schulhofer, supra note 184, at 468-69. See also DRESSLER, supra note 4, at §32.02.
192 See Schulhofer, supra note 184, at 468-69.
193 See Steinglass, supra note 35, at 361.
194 See Schulhofer, supra note 184, at 487.
195 See DRESSLER, supra note 4, at 607.
196 See Schulhofer, supra note 184, at 454 (noting that “…reprosecution may be barred even though
no adjudication results from the first proceeding. The doctrine thus provides more meaningful protection
against the danger of governmental harassment and the burden of repeated trials…”). Schulhofer also
makes the point that “A number of courts have barred retrial even when mistrial was triggered by absence
of the defendant, impermissible cross-examination, or persistently objectionable behavior by defense
counsel.” See Schulhofer, supra note 184, at 484.
197 See generally Cheh, infra note 281. See also Mann, infra note 281, at 1869-73.
198 See United States v. Halper, 490 U.S. 435, 449 (1989); U.S. v. Ward, 448 U.S. 242, 248-249 (1980).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 47
and (7) whether it appears excessive in relation to the alternative
purpose assigned to it.199
Many of these factors appear to be correlated with concerns about rent-
seeking. For example, consider the concern with the magnitude of the sanction
(7) or whether it involves an affirmative disability (1). As the civil sanction
becomes larger in magnitude (either in terms of monetary or non-monetary
sanctions) then the ability of the prosecutor to extract wealth increases.200 This is
because the prosecutor has a bigger threat to hold over the defendant as sanction
severity increases. The other factors may also represent concerns with reducing
rent-seeking and appear consistent with our approach.201 Double Jeopardy,
however, is not the only area of criminal procedure that appears to reflect rent-
seeking concerns. 202
B. Ex Post Facto Clause
The ex post facto clause bars retroactive application of certain changes in
the criminal law. 203 The standard justifications for this rule are that it provides
notice to defendants about what is illegal and the sanction for it, as well as
constraining the government from passing arbitrary or vindictive legislation
against a particular defendant.204 The prohibition is only concerned with matters
that amount to “punishment”205 and applies more frequently in the context of
legislative decisions as compared to judicial decisions. 206
199 Hudson v. United States, 522 U.S. 93, 99 (1997); Kennedy v. Mendoza-Martinez, 372 U.S. 144, 168
- 169 (1963).
200 See Friedman, supra note 18, at 268.
201 Some of the other factors (e.g., scienter, historically regarded as punishment) may work as
proxies for the kind of stigma associated with the particular civil wrong. See Cheh, infra note 281, at 1352-54.
If the wrong appears like what most people consider criminal then the stigma may be quite high thereby
giving the prosecutor greater ability to extract wealth. The fourth and fifth factors suggest that one
important purpose for the doctrine in this area is to discourage enforcement agents from substituting civil
enforcement for criminal enforcement in order to evade the procedural protections that come along with
criminal enforcement. See id. at 1345, 1354-57, 1394.
202 We should note that we are not claiming that every aspect of Double Jeopardy case law (which
is not generally considered a model of clarity and consistency) matches up with a rent-seeking theory. See
Albernaz v. United States, 450 U.S. 333, 343 (1981) (noting that the Double Jeopardy decisional law is "a
veritable Sargasso Sea which could not fail to challenge the most intrepid judicial navigator"). However, we
do suggest that some important parts of Double Jeopardy seem to display a concern with rent-seeking.
Second, simply because Double Jeopardy may address concerns of prosecutorial abuse does necessarily
mean that it is the best place to address those concerns. Many of these concerns, especially the mistrial
context, might be better addressed under other parts of the Constitution (e.g., Due Process), but we do not
make any comment on that issue at this time. See Akhil Amar, Double Jeopardy Law Made Simple, 106 YALE L.J.
1807, 1809 (1997).
203 See JOHN E. NOWAK & RONALD D. ROTUNDA, CONSTITUTIONAL LAW 428 (5th ed. 1995). See also
Lindsey v. Washington, 301 U.S. 397 (1937).
204 See WAYNE R. LAFAVE & AUSTIN W. SCOTT, JR. CRIMINAL LAW 97 - 100 (2d. ed).
205 The term “punishment” is not the same as the label “criminal”. Sometimes an act labeled “civil”
by the legislature will nonetheless amount to “punishment” for purposes of ex post facto inquiry. See
Landgraf v. U.S.I. Film Products, 511 U.S. 244 (1994) (finding that the punitive damage awards of the 1991
Toward an Economic Theory of Pro-Defendant Criminal Procedure 48
Traditionally the ex post facto rule applies in the four contexts set out in
Calder v. Bull.207 These are: (1) when the legislature creates a new criminal law it
may not be applied retroactively to behavior that was not criminal at the time it
occurred; (2) when the legislature removes elements from the definition of a
crime or otherwise increases the severity of a crime (making it a more serious
crime than when it occurred) these changes may not be applied retroactively; (3)
when the legislature increases the punishment for a particular crime then this
change may not be applied retroactively; and (4) when the legislature changes
the rules of evidence or changes the requirements for testimony relative to what
they were when the act occurred then the changes may not be applied
retroactively.208 Contexts (2) and (3) are often considered as one.209
Context (1) is seldom brought into question in modern times but it still
represents the quintessential instance of the ex post facto prohibition.210 It
represents an obvious instance of where the government and the prosecutor are
abusing their discretion and targeting a particular defendant probably in
response to some perceived political, or other, gain they may receive. 211 This
seems to fit easily within our approach to procedural protections – constraining
the costs associated with abuse of discretion.
Contexts (2) and (3) seem to still arise in some form in modern times.212
They also represent instances, like Context (1), where the concern with
Civil Rights Act were similar enough to criminal sanctions to apply the ex post facto clause). However, this is
not a frequent occurrence so that normally the legislative label is determinative (i.e., if it is labeled “civil” by
the legislature then it will most likely not amount to “punishment”). See Kansas v. Hendricks, 521 U.S. 346
(1997).
206 This is because the prohibition appears in Art. I of the U.S. Constitution, which deals with
legislative power and not in Art. III, which deals with judicial power. See LAFAVE & SCOTT, supra note 204, at
97 - 100.
207 See Calder v. Bull, 3 U.S. 386 (1978).
208 See id. at 390.
209 See LAFAVE & SCOTT, supra note 204, at 97 n.3; NOWAK & ROTUNDA, supra note 203, at 428.
210 See LAFAVE & SCOTT, supra note 204, at 98.
211 See David Friedman, Making Sense of English Law Enforcement in the 18th Century, 2 U. CHI. L.
SCH. ROUNDTABLE 475 (1995).
212 Dobbert v. Florida, 432 U.S. 282 (1977) Malloy v. South Carolina, 237 U.S. 180 (1915); Hernandez
v. State, 43 Ariz. 424 (1934); Lindsey v. Washington, 301 U.S. 397 (1937). The ex post facto prohibition will
also be violated if a statute eliminated a former element of the offense or took away a defense that was
formerly available. See Beazell v. Ohio, 269 U.S. 167 (1925); Kring v. Missouri, 107 U.S. 221 (1882) (noting
that “the new Constitution of Missouri does take away what, [by] the law of the State when the crime was
committed, was a good defence to the charge of murder in the first degree.”); Thompson v. Utah, 170 U.S.
343 (1898) (arguing that “the provision in the constitution of Utah providing for the trial in courts of general
jurisdiction of criminal cases, not capital, by a jury composed of eight persons, is ex post facto in its
application to felonies committed before the Territory became a State, because, in respect of such crimes, the
Constitution of the United States gave the accused, at the time of the commission of his offence, the right to
be tried by a jury of twelve persons, and made it impossible to deprive him of his liberty except by the
unanimous verdict of such a jury.”).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 49
prosecutors abusing their discretion and engaging in selective enforcement is
significant.213 Thus, these contexts also fit with our approach.
A related issue is what happens when the government passes a law that
changes something related to parole requirements or privileges.214 In this context
the courts have adopted something of a sliding scale approach. In certain cases
the changes, when applied retroactively, may violate the ex post facto clause,
whereas in other cases the changes may not violate the clause even when applied
retroactively.215 We think our theory provides the best explanation for these
cases.
Consider a fairly recent US Supreme Court decision – California Dept. of
Corrections v. Morales.216 Morales involved a case where the California State
government changed the rule (contained in a statute) on reconsideration
hearings for prisoners who had their first attempt at obtaining parole rejected. 217
The old rules provided for annual reconsideration for such inmates and the new
rules provided for the parole board to defer hearings for up to 3 years if the
board “finds that it is not reasonable to expect that parole would be granted…
during the following two years and states the basis for its findings”. 218 In other
words the new rule changed the statutory scheme and granted additional power
and discretion to the parole board. The Court held that this rule, when applied
retroactively, did not violate ex post facto because there was only a remote
likelihood of obtaining parole in those cases where the new rule applied (i.e., the
new rule required parole board to believe that there was no reasonable chance to
receive parole before deferring the hearings).219 Thus, there was only a
speculative possibility of really extending the prison term through this new rule
(as there was little chance parole would be granted even if an earlier hearing had
213 In the older cases any change (whether to increase or decrease the punishment) might have
violated ex post facto if applied retroactively. See In Re Tyson, 13 Colo. 482 (1889); Commonwealth v.
McDonough, 95 Mass 581 (1866). However, now that has changed so that a decrease in punishment is
unlikely to violate ex post facto. See U.S. v. Stewart, 1993 U.S. App. Lexis 17634, *10, 1993 WL 265147, **3
(10th Cir. 1993) (noting that “the framers of the Ex Post Facto Clause intended it to preclude only increased
punishment for preexisting criminal conduct”). See also Miller v. Florida, 428 U.S. 423, 430 (1987). The older
cases are consistent with our approach because they deter potential defendants from lobbying to increase
sentences for other groups and from lobbying to decrease sentences for themselves. The more recent cases,
however, seem to provide for asymmetric protection. See U.S. v. Stewart, supra. That is, they protect against
lobbying that increases sentences for a particular group, but do not stop parties from lobbying against
decreases in sentences for themselves. This may be desirable because either this form of rent-seeking is
addressed elsewhere (say in anti-corruption statutes) or because a rule that banned ex post facto decreases
in sentences may not stop lobbying much as the defendant would simply shift their efforts to lobbying the
prosecutor not to bring a case – a matter that cannot be reviewed in court.
214 See 16A Am. Jur. 2d, Constitutional Law §643.
215 See Garner v. Jones, 120 S.Ct. 1362 (2000); California Dept. of Corrections v. Morales, 514 U.S. 499
(1995); Weaver v. Graham, 450 U.S. 24 (1981).
216 See California Dept. of Corrections v. Morales, 514 U.S. 499 (1995).
217 See Morales, supra note 216, at 501.
218 See id. at 503.
219 See id. at 513.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 50
occurred).220 This suggests little room for abuse of discretion or increase in
discretion because the Parole board would probably have denied parole anyway
so that delaying the hearings was not likely to significantly increase the risk of
the defendant spending a longer period of time in jail.221 Given this small risk
the concerns behind the ex post facto clause did not seem triggered.
This approach matches nicely with ours, which is a concern with the
potential for rent-seeking activities following from an increase in
prosecutorial/parole board discretion. In Morales although there was what
appeared to be an increase in discretion it was fairly well constrained. This is
because the board could only defer hearings for those prisoners that had no
reasonable chance of parole. If a prisoner had little chance of parole anyway
then the scope for abusing discretion should be quite limited and the concerns
with rent-seeking more muted.222
Context (4) involves instances where changes in the standard of proof and
rules of evidence are applied retroactively. This is prohibited by the ex post facto
clause.223 One of the older English cases establishing this point involved a case
where the defendant had convinced a co-conspirator in treason to flee the
country.224 The English Parliament responded by reducing the number of
witnesses needed to convict for treason, which the court held to violate the ex
post facto prohibition.225 This case fits neatly into our approach as it involves an
instance where the legislature changes the law in order to target a particular
defendant.
More recent case law has adopted a more nuanced approach. For
example, the new case law distinguishes between rules that affect witness
competency and rules that change the quantum of evidence needed to convict
220 See id. at 509, 514. The Parole board was required to provide reasons for its decisions too. See
Morales, supra note 216, at 511.
221 See id. at 512.
222 The US Supreme Court has recently upheld the Morales decision in Garner v. Jones, 120 S.Ct.
1362 (2000). The facts in Garner are similar to Morales in that the parole board was able to delay hearings on
parole as a result of a regulatory amendment (not a statutory change as in Morales). See Garner, supra, at 1366
– 69. The Court remanded and required a finding on whether there was a “significant risk” under Morales.
See id., at 1370. There are some differences between Garner and Morales. In Garner the regulatory
amendment applied to a broader class of prisoners than in Morales and in Garner the amendment did not,
arguably, change the parole board’s discretion – both before and after the amendment the parole board had
complete discretion to grant parole by statute. See id., at 1369 – 71. The change only helped guide that
discretion. See id., at 1369. In Morales, the statutory change did increase board discretion. See Morales, supra
note 216, at 507. This difference may be a reason not to apply the ex post facto prohibition here (as there
may not appear to be an increase in discretion on the facts of Garner). This seems the implicit approach of
Justice Scalia in his concurrence. See Garner, supra, at 1371 (Scalia, J. concurring).
223 See Walker v. State, 433 So. 2d 469 (Ala. 1983); Plachy v. State, 239 S.W. 979 (1922); Thompson v.
Missouri, 171 U.S. 380 (1898); LAFAVE & SCOTT, supra note 204, at 97 – 101.
224 See 9 T. MACAULAY, HISTORY OF ENGLAND, 31, 171-173 (1899).
225 See id. at 270.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 51
defendants. In Carmell v. Texas,226 the US Supreme Court held that a change in
the law regarding when uncorroborated testimony of sexual assault
complainants would be admitted violates the ex post facto clause if applied
retroactively.227 Prior to the change in the law, Texas required that the testimony
of sexual assault complainants be corroborated unless the complainant was
under the age of 14.228 After the defendant’s alleged wrongdoing the law
changed so that uncorroborated testimony of any complainant below 18 years of
age would be acceptable. 229 There appeared to be no evidence that this change
occurred in response to any particular defendant before the courts at the time. 230
The prosecution, however, attempted to use this law here because during the
relevant time the complainant was between the ages of 14 and 18.231 The court
held that retroactively using this law violated the ex post facto prohibition
because it reduced the quantum of evidence necessary to convict the defendant,
which the majority compared to being as oppressive as changing the
requirements for the offense. 232
The dissent argued that the majority’s decision ran counter to Hopt v.
Territory of Utah,233 where the Supreme Court held that a change in the law that
permitted the prosecution to present the evidence of witnesses convicted of
felonies was not a violation of ex post facto even if applied retroactively.234 The
dissent viewed the Hopt case as being indistinguishable from the case at hand
and would not have barred the retroactive use of the law.235 The majority
addressed this issue by saying that the Hopt decision was about witness-
competency statutes, whereas the Carmell case was about the quantum of
evidence needed to convict the defendant. 236 Indeed, the Hopt decision itself
made a distinction between the facts in that case and cases where the quantum of
evidence required for conviction had changed.237
Can such a distinction be justified? We think this distinction may have
some force under our approach. If our concern is with the ability of the
prosecutor to use certain changes in the law of evidence to increase the chance of
abuse of the criminal process then the Hopt and Carmell contexts present differing
risks of this abuse.
226 See Carmell v. Texas, 120 S. Ct. 1620 (2000).
227 See id. at 1643.
228 See id. at 1624.
229 See id. at 1625.
230 See Carmell, supra note 226, at 1651 (Ginsburg, J., dissenting).
231 See Carmell, supra note 226, at 1625.
232 See id. at 1633.
233 See Hopt v. Territory of Utah, 110 U.S. 574 (1884).
234 See id. at 589.
235 See Carmell, supra note 226, (Ginsburg, J., dissenting) at 1643, 1653, 1655.
236 See id., at 1639.
237 See Hopt, supra note 233, at 590.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 52
In Hopt the prosecution is being permitted to use testimony of those who
have committed felonies.238 Given that a witness’s prior record may be used in
court to challenge a witness’s testimony it is not clear how permitting felons to
testify greatly increases prosecutorial discretion.239 The jury or the judge may
already view a felon’s testimony with some skepticism so, although there is some
room for abuse, it seems unlikely that there would be great room for
prosecutorial abuse across most cases relative to the Carmell context. 240
In the Carmell context (a sexual assault), obtaining the testimony of the
complainant is probably quite important to the case. Increasing the prosecutor’s
discretion with regard to this type of testimony (which may generally have
greater credence and importance to a jury or judge than the testimony of a non-
complainant felon) gives the prosecutor and other parties considerably greater
room to use the system to their advantage.241 Thus, the change in Carmell may
make it significantly easier to convict (or credibly threaten to convict) a
defendant, whereas the change in Hopt may only slightly increase this risk
because the testimony of felons may not generally carry great weight in many
cases.242 In other words, the potential for abuse is greater (and hence the costs
associated with rent-seeking higher) in the Carmell context compared to the Hopt
context.243
C. Some Other Measures That Constrain Rent-Seeking
There are many other doctrines, in addition to those discussed so far, that
constrain rent-seeking in the criminal law enforcement process. In this Part we
address two of them briefly, void-for-vagueness doctrine and entrapment.
238 See id. at 587.
239 See id. at 588.
240 One could, of course, posit instances where such abuse may occur (e.g., the prosecution offering
a felon, who is currently in jail, a reduced sanction in some manner for fabricating testimony), but one
suspects that the risk is either not great or that the testimony would not be generally believed. See Joshua M.
Levinson & Brian Lambert, Twenty-Ninth Annual Review of Criminal Procedure, 88 GEO. L.J. 1175 (2000)
(discussing government’s disclosure obligations). Note that the Hopt decision concerned a law that related
to permitting all felons (even those not in jail and hence not subject to as much prosecutorial arm-twisting)
so that the threat of prosecutor’s using their power against this general group is not entirely persuasive. See
Hopt, supra note 233, at 588. The facts of Hopt involved a felon in prison at the time, but the rule was not
limited to those instances. See id. at 589.
241 See Carmell, supra note 226, at 1640.
242 See id.
243 Note that we have only argued that the potential for abuse is greater in Carmell than Hopt not
necessarily that Hopt was correctly decided – for that to be the case we would need to believe that the
potential for abuse in Hopt was below the threshold, whatever it might be, that is needed to trigger ex post
facto prohibitions. We make no comment on that at this stage except to argue that there is a difference in
abuse potentials between the cases. Also we make no comment on where the threshold for triggering ex
post facto prohibitions should/might be.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 53
1. Void-for-vagueness
The void-for-vagueness doctrine serves as a constraint on legislators and
law enforcement agents that curtails their discretion. The early cases struck
down laws that were deemed vague,244 in the sense that they granted law
enforcement agents broad discretion in deciding what is legal and what is not.245
Such discretion gives enforcement agents wide power to extract wealth through
the criminal law enforcement process. 246 This raises the specter of large rent-
seeking costs.
More recently a new twist to the issue of vague statutes has arisen: what
happens when the group against which selective enforcement may be used
agrees to it or supports it? This is the situation that gave rise to the Supreme
Court’s decision in Chicago v. Morales.247 The case developed after Chicago
passed an ordinance that prevented “criminal street gang members from
loitering with one another or with other persons in any public place.” 248 The
definition of loitering was quite broad – “[remaining] in any one place with no
apparent purpose”249 and police were required to ascertain whether some (at
least one of two) persons who were “loitering” were gang members.250 If so, the
244 See Connally v. General Constr. Co., 269 U.S. 385 (1926) (stating on the facts that “[t]he term
‘current rate of wages’ referred to minimum, maximum, and intermediate amounts, thus the term was too
vague for appellee to know to which amount the statute referred. In addition, the term ‘locality’ had no
precise meaning. Thus, the statute did not allow employers to know what the minimum wage was, and was
therefore unconstitutionally vague”); Miller v. Schone, 276 U.S. 272 (1928) (noting that “the statute is void
for vagueness and uncertainty. It contains no criterion whatever by which to determine who are the
freeholders of the locality to whom is confided the power of invoking the axe of the Entomologist. Again,
what is the ‘locality’ intended by the statute? No technical meaning attaches to the term”).
245 See Michael K. Browne, Current Public Law and Policy Issues: Loitering Laws: Does Being “Tough On
Crime” Justify the City of Minneapolis’ Use Of a Vague and Broadly Constructed Ordinance, Which Criminalizes
Out Thoughts in Violation of the First Amendment?, 20 HAMLINE J. PUB. L. & POL. 147, 148 (1998) (arguing that
“an officer's hunch becomes the basis for the suspected activity, and charges are brought against
‘undesirable’ individuals without probable cause that a criminal act has been performed. Somehow, officers
and prosecutors systematically determine that minorities are the ‘undesirables’ and they perpetuate
unconstitutional arrests based upon this faulty ordinance.”); Erik Luna, Transparent Policing, 85 IOWA L. REV.
1107, 1132 (2000) (noting that “as a matter of history and practical necessity, police and prosecutors are
vested with broad latitude in their application of the penal code--to detain, interrogate, and arrest suspects,
for instance, or to charge and prosecute defendants in the criminal process. Sometimes these discretionary
powers are openly admitted, conspicuously employed, and, thereby, exposed to popular review.”).
246 See Lars Noah, Administrative Arm-Twisting In the Shadow of Congressional Delegations of Authority,
1997 WIS. L. REV. 873, 903 (1997) (arguing that “criminal defendants routinely plead guilty, typically in
exchange for a reduction in charges or some concession by prosecutors in making sentencing
recommendations. An accused may agree to plead guilty to a lesser-included offense or a smaller set of
offenses charged in an indictment if the prosecutor agrees not to pursue other charges in the indictment.
Indeed, defendants may avoid prosecution altogether by agreeing to participate in a pretrial ‘diversion’
program, such as drug rehabilitation for certain types of offenders.”).
247 See Chicago v. Morales, 527 U.S. 41 (1999).
248 Id. at 45 – 46.
249 Id. at 47 & n.2.
250 Id. at 47 & n.2.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 54
police could order them to leave the area. 251 The ordinance, arguably, had the
support of the community in which it was most likely to be enforced, the high-
crime urban neighborhoods of Chicago.252 It was argued that these
neighborhoods supported the ordinance in order to gain greater safety within
their communities.253 In spite of the community support, the Supreme Court
held the ordinance unenforceable on vagueness grounds.254 The Court’s reasons
included that the ordinance defined “loitering” and other matters in such a broad
way that they were impossible to obey.255
In light of the broad grant of discretion to the police under Chicago’s anti-
gang ordinance it is obviously possible that the police could engage in selective
enforcement.256 The community may have consented to this risk in order to
enhance its security.257 The general question raised by Morales is whether a
community should be allowed to make this trade off.
Some scholars have argued for an exception to the void-for-vagueness
doctrine on the ground that selective enforcement or targeting is extremely
unlikely in the Morales context.258 Specifically, the use of an anti-gang ordinance
to oppress a particular group, or transfer wealth from one group to another, is
extremely unlikely in the Morales setting for two reasons. First, the ordinance,
arguably, had a high degree of community support even within the
neighborhoods most likely to be burdened by its enforcement.259 Second, the
costs of selective enforcement probably would have been internalized within the
relevant communities.260 Put another way, this case is unlike the example of
selective law enforcement in the Jim-Crow South, which involved the use (or
non-use) of force by a politically dominant group to oppress a politically
marginal group. The communities that supported the anti-gang ordinance made
an apparently conscious decision to trade off protection from police harassment
in order to reduce the crime rate in their neighborhoods. The general implication
251 Id. at 47.
252 See Morales, supra note 247, at 74 (Scalia, J., dissenting) (noting that “[m]any residents of the
inner city felt that they were prisoners in their own homes...Chicagoans decided that to eliminate the
problem it was worth restricting some of the freedom they once enjoyed.”).
253 See id., at 51, 74. See also Tracey L. Meares & Dan M. Kahan, The Wages of Antiquated Procedural
Thinking: A Critique of Chicago v. Morales, 1998 U. CHI. LEGAL F. 197, 197-213.
254 See Morales, supra note 247, at 51.
255 See id., at 56 – 59.
256 See id., at 58 – 59.
257 See id., at 74 (Scalia, J., dissenting) (stating that “[t]he minor limitation upon the free state of
nature that is the prophylactic arrangement imposed upon all Chicagoans seemed to them (and it seems to
me) a small price to pay for liberation of their streets”).
258 See Meares & Kahan, supra note 253.
259 See id.; Morales, supra note 247, at 74 (Scalia, J., dissenting) (stating that “[t]he minor limitation
upon the free state of nature that is the prophylactic arrangement imposed upon all Chicagoans seemed to
them (and it seems to me) a small price to pay for liberation of their streets”).
260 This seems implicit in Justice Scalia’s approach. See Morales, supra note 247, at 74. For greater
discussion see Brooks, infra note 262.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 55
of this argument is that constitutional restraints on criminal law enforcement
should be relaxed in settings where the risk of selective enforcement is minimal.
For simplicity, we will refer to this as the internalization critique.
The framework of this paper provides an alternative to the internalization
critique, as well as a justification for the Court’s decision in Morales. The
internalization critique misses an important feature of pro-defendant procedural
protections. Not only do they dampen temptation for selective enforcement, or
inter-group wealth expropriation, they also cut down the prospects for what we
have termed “simple corruption.” That is, procedural protections also have the
function of reducing the opportunities for an individual enforcement agent to
enrich himself by using his position to bully individuals who can be threatened
with arbitrary arrest and punishment. This is potentially just as harmful as inter-
group wealth extraction, because as long as it is possible for individuals to enrich
themselves through the enforcement process, people will devote resources to
acquiring positions as enforcement agents. Each position along the chain of
enforcement (from the officer on the street to his immediate superiors, to
prosecutors, to parole officers) could become a source of monopoly profits for the
individuals who occupy them. 261 When this occurs on a large scale, consistency
and impartiality in enforcement are unlikely to be observed.262
2. Entrapment
The fact that the entrapment defense is a relatively new common law
doctrine probably has a lot to do with the expanding scope of criminal
prohibitions.263 The defense does not exist for common law crimes, such as
261 This is essentially the argument made in the context of a vertically fragmented enforcement
scheme. See supra text accompanying notes 105 – 107.
262 In addition to this argument it should be noted that the factual predicates of the case – that the
minority dominated community supported this measure and hence was willing to trade off civil rights for
enhanced safety – is a fairly contentious matter. First, there is significant debate over whether the
community did actually support these measures. See Richard R.W. Brooks, Fear and Fairness in the City:
Criminal Enforcement and Perceptions of Fairness in Minority Communities, 73 S. CAL. L. REV. 1219 (2000). The
support seems to be equivocal. See id., at 1233-35. Second, even if the community did support this measure
it does not tell us too much about how much the community is willing to trade rights for safety. See id., at
1262 (noting that “…if poor blacks are more supportive of the American legal system because they are less
aware of the existence of race-based unfairness…, then a desire or willingness on their part to expand legal
enforcement in poor urban communities is not a fully informed position for lawmakers to follow.”). All it
says is (assuming the community did support the measure) that the community preferred this mix of safety
and civil rights over the current one. It does not tell us that this option would have been preferred over
others that were not offered to the community. Indeed, it is possible that other alternatives could have been
preferred by the community to the ordinance or the current state of affairs. Community support therefore
only tells us so much. See id. at 1271. Finally, even if the community did support this ordinance over all
others that does little to address concerns with the prosecutors’ now enhanced power to extract wealth.
263 See Sorrells v. United States, 287 U.S. 435, 453 (1932) (Roberts, J., concurring) (noting that “[t]he
increasing frequency of the assertion that defendant was trapped is doubtless due to the creation by statute
of many new crimes, (e.g., sale and transportation of ...narcotics) and the correlative establishment of special
enforcement bodies for the detection and punishment of offenders”).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 56
murder. 264 This is a sensible result because even those who commit murder as a
result of inducement, cajolery, or solicitation are still likely to present a danger to
society.265 The defense is connected today largely with drug prosecutions and
other victimless crimes.266
Our theory provides a rationale for the entrapment defense and for its
relatively recent appearance in the law. Entrapment’s recent appearance in
connection with new prohibitions can be understood as a reaction to the rent-
seeking hazards associated with expanding criminal prohibitions. The short list
of common law crimes encompasses conduct that is uniformly considered
undesirable.267 It is possible, as we have argued, for enforcement agents to
enforce them selectively, but there are many procedural protections in existence
to constrain this incentive. Relatively new criminal prohibitions, on the other
hand, often encompass conduct that is not uniformly considered undesirable,
and may quite easily be made the basis for selective enforcement.268 Consider,
for example, the debates concerning the more severe punishments for crack
cocaine, heavily used in minority neighborhoods, relative to powdered
cocaine.269 As a positive matter, then, we should expect to observe, and we have
observed, the entrapment defense expanding in scope and gaining a stronger
footing in criminal law doctrine as the scope of criminal prohibitions extends
beyond basic common law crimes.
One rationale provided for the entrapment defense is that it enables courts
to avoid becoming tainted by condoning inappropriate conduct, or “abhorrent
264 See LAFAVE & SCOTT, supra note 204, at 421 – 22; See also MODEL PENAL CODE §2.13 (noting that the
defense of entrapment is unavailable when “causing or threatening bodily injury is an element of the
offense”).
265 See MODEL PENAL CODE §2.13 Comment at 420 (1985) (noting that “one who can be persuaded to
cause such injury presents a danger that the public cannot safely disregard”).
266 See Dana M. Todd, In Defense of the Outrageous Government Conduct Defense in the Federal Courts,
84 KY. L. REV. 415, 419 (1995); John F. Pries, Witch Doctors and Battleship Stalkers: The Edges of Exculpation in
Entrapment Cases, 52 VAND. L. REV. 1869, 1872 (1999).
267 These common-law crimes are mala in se, or morally wrong acts. These crimes are distinguished
from mala prohibita crimes, which are acts made criminal by statute, but are not of themselves considered
criminal. See BLACK’S LAW DICTIONARY 956 (6th ed. 1990). One argument in support of excluding the
entrapment defense from conduct uniformly considered criminal is that, “from a moral perspective, it is
wrong to punish those…who lack an opportunity to know and adhere to the law due to government
conduct.” John T. Parry, Culpability, Mistake and Official Interpretations of Law, 25 AM. J. CRIM. L. 1, 5 – 6
(1997).
268 See Sorrells, supra note 263, at 453 (Roberts, J., concurring) (noting that “efforts…to obtain arrests
and convictions (of these crimes) have too often been marked by reprehensible methods.”). See also Sanford
H. Kadish, Fifty Years of Criminal Law: An Opinionated Review, 87 CALIF. L. REV. 943, 970 (1999).
269 See United States v. Anderson, 82 F.3d 436 (D.C. Cir 1996); United States v. Sanchez, 81 F. 3d 9
(1st Cir. 1996) (holding that increased sentencing for possession of crack rather that powdered cocaine was
not unconstitutionally void for vagueness). See also Stuntz, supra note 76; Michael R. Bromwich, Put a Stop to
Savage Sentencing, WASH. POST, Nov. 22, 1999, at A23 (discussing the unfairness of the disparity in
sentencing).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 57
transaction(s)”.270 The rationales could be expanded to include the claim that it
discourages police officers from engaging in inappropriate conduct because it
effectively denies them the reward (in terms of prosecutions) for engaging in
such conduct.271 This rationale has been criticized as inadequate on the ground
that the purity of the courts, or of enforcement agents, has no particular value in
itself.272 If some impurity enhances the deterrent effect of the law, why not allow
it?
We think our framework provides a stronger rationale for the entrapment
defense. The function of the entrapment defense is not, in our view, to simply
protect the purity of enforcement agents, but to dampen rent-seeking incentives
at lower levels of the enforcement process. If enforcement agents are denied the
fruit of entrapment efforts, the rewards from using the law enforcement process
to target specific individuals or groups fall.
D. Some Empirical Evidence
In addition to providing a better rationale for pro-defendant criminal
procedure doctrines, the rent-seeking framework of this paper is also
corroborated by corruption evidence from several countries. We ran a regression
of Transparency International’s corruption index on several variables, including
measures of key pro-defendant criminal procedural rules. The reasoning behind
this exercise is that if pro-defendant criminal procedural rules reduce the
incentives to use the criminal laws for inter-group wealth extraction and for
personal enrichment, the degree of corruption should be lower in countries that
have such procedural rules.
The key measures of pro-defendant rules used in the regression analysis
are two: the existence of a rule prohibiting cruel and unusual punishment, and
the existence of a reasonable-doubt standard. The results suggest that both types
of pro-defendant protection are strongly negatively correlated with the degree of
corruption. In other words, the presence of these protections is strongly
correlated with lesser corruption.
Transparency International’s corruption index provides a score ranging
from 10 (least corrupt) to 1 (most corrupt) for roughly 80 countries. The index,
which measures international perceptions of corruption (bribe-taking and bribe-
270 Sorrells, supra note 263, at 459 (Roberts, J., concurring).
271 See id. at 448.
272 The majorities in Sorells and Sherman used a subjective test in applying the entrapment defense
(focusing on the defendant’s predisposition for crime and mental state), rather than focusing on the action of
government and law enforcement officials. See Sorrells, supra note 263, at 451; Sherman v. United States, 356
U.S. 369, 375 – 76 (1958). See also Jason R. Schulze, United States v. Tucker: Can the Sixth Circuit Really Abolish
the Outrageous Government Conduct Defense?, 45 DEPAUL L. REV. 943 (1996).
Toward an Economic Theory of Pro-Defendant Criminal Procedure 58
paying), is based on a survey of business people and analysts.273 Our measure
for the existence of a reasonable-doubt standard is simply coding for whether the
country has a common law system. In general, the reasonable-doubt standard
seems to be a feature largely of common law legal regimes.274 The cruel and
unusual punishment measurement is reflected by two “dummy variables.” One
variable, Crupun3 takes the value one if the country either does not have a rule
prohibiting cruel and unusual punishment, or does not abide by the rule if it has
one. The other variable, Crupun2 takes the value one if the country has a rule
prohibiting cruel and unusual punishment, but there are some concerns
expressed about the state’s compliance with its own rule.
We also included in the regression a measure of the ratio of government
spending to gross domestic product (GDP). The purpose of this measure, labeled
Ratio, is to capture the relative number of opportunities for corruption in a
country. Presumably as this ratio increases, the number of enforcement agents
increases relative to the size of the economy. For example, a country that has one
licensing agent for every business will presumably have a large ratio of
government spending to GDP. It happens, however, that this argument is
inadequate because a country may choose to pay its licensing agents nothing
(allowing them to make up the shortfall in bribes) and then the ratio of
government spending to GDP may be relatively small. To take this into account
we ran a second version of the regression, taking into account the relative rates of
private and public sector pay.
The results in Table 1 are for a regression of the Corruption index (labeled
CPI) on Ratio and the pro-defendant procedure variables. The results indicate
that corruption is significantly lower where the reasonable-doubt rule is in effect,
as measured by the common law variable.275 Moreover, both measures of cruel
273 For information on the Corruption Perceptions Index, see
. See also Johann Graf Lambsdorff, Background Paper to
the 2000 Corruption Perceptions Index (September 2000).
274 See CRAIG M. BRADLEY, CRIMINAL PROCEDURE: A WORLDWIDE STUDY xv-xxii (1999), for a
discussion of the major differences in criminal procedure between common law and civil law systems. In
civil law, or inquisitorial, systems (found in a majority of continental European countries) a “theoretically
neutral judicial officer conducts the criminal investigation and a judge…determines guilt or innocence.” Id.
at xv. Common law systems (found in the United States, Great Britain and its former colonies) are based on
a mistrust of the government, and “the defendant is endowed with a quiver of rights that he may launch
against the government at various stages of the proceeding.” Id. at xvi. The reasonable-doubt standard is
one of the defendant’s weapons against common law criminal systems. England, Wales, South Africa, and
the United States all require that the defendant’s guilt be shown beyond a reasonable doubt. See id. at 122,
349.
275 The results in Table 1 were largely replicated in a second regression that includes a variable
measuring the ratio of public sector wages to financial sector wages. In the second regression, the
COMMLAW and CRUPUN2 coefficients remained roughly the same. The CRUPUN3 variable dropped to
statistical insignificance, but this may largely be a byproduct of the sharp drop in observations because of
missing wage date. We had only 42 observations for the second regression. The new variable PAFIN,
Toward an Economic Theory of Pro-Defendant Criminal Procedure 59
punishment constraints indicate that the failure to prohibit such punishment is
positively correlated with corruption. The results indicate that moving from a
regime in which there is a prohibition of cruel and unusual punishment (that is
complied with) to one in which there is no such prohibition (Crupun3) reduces
the Corruption index (i.e., increases corruption) by 3 points. This is quite a
substantial drop given that the maximum score is 10. The existence of a common
law system raises the Corruption index (i.e., reduces corruption) by 1.5 points.
Table 1
CPI Coef. Std. Err. t P>|t|
RATIO .039 .022 1.800 0.076
CRUPUN3 -2.997 .513 -5.840 0.000
CRUPUN2 -2.104 .604 -3.482 0.001
COMMLAW 1.455 .471 3.090 0.003
CONS 4.986 .730 6.832 0.000
Number of obs = 75
R-squared = 0.458
Adj R-squared = 0.427
The substantial impact of the two variables measuring cruel punishment
constraints and the common law variable were replicated in expanded regression
models controlling for educational levels (percentage at primary level), religion
(percent catholic, muslim), and economy type (socialist, mixed).276 Although the
coefficient for Crupun3 fell in absolute value from 3 to 2, it remained statistically
significant and increased in proportion to the common law variable.
Interestingly, the results suggest that the cruel and unusual punishment
measures have a much larger impact on corruption than the common-law
measure (which proxies for the reasonable-doubt standard). This has interesting
implications for the recent literature on common law protections and economic
growth.277 The results suggest that hard constraints on the state’s freedom to
which measures the ratio of public sector to financial sector wages, came in highly significant with a
coefficient of 3.395 (t-statistic = 2.3).
We also ran the same regression with controls for education, religion, and economy type (socialist
versus capitalist), and the coefficients on COMMLAW, CRUPUN2 and CRUPUN3 remain roughly the same.
276 See Appendix.
277 For greater discussion see, e.g., Rafael LaPorta, Florencio Lopez-de-Silanes, Andrei Schleifer &
Robert W. Vishny, Law and Finance, 106 J. POL. ECON. 1113, 1151 – 52 (1998); Paul G. Mahoney, The Common
Law and Economic Growth: Hayek Might be Right, University of Virginia School of Law, Legal Studies Working
Paper Series, Working Paper 00-8, January 2000, available at
Toward an Economic Theory of Pro-Defendant Criminal Procedure 60
profit through punishment may be a more important restriction on corruption
than the existence of common law rules.
VIII. CONCLUSION
The strong pro-defendant bias in Criminal Procedure is a stalwart of
Anglo-American jurisprudence. This has often seemed perplexing because it
may increase the incidence of criminal wrongdoing, whereas one of the primary
reasons for declaring something “criminal” is to try to reduce its incidence.278
Such an apparent contradiction has led to many attempts to justify this approach
to the criminal process. Our paper provides a simple economic explanation for
why such a strong pro-defendant bias may be justified – to constrain the costs
associated with self-interested behavior by prosecutors and government agents -
and finds that case law and certain empirical evidence is consistent with our
theory.
We begin by sketching some of the more common criminal procedures
and examining the traditional justifications given for them. In particular, we
focus on the reasonable doubt standard of proof and the justification most
commonly given for it – that we are more concerned with the costs associated
with false convictions than with false acquittals. We find it highly unlikely that
the traditional rationale provides a good justification for the pro-defendant bias
in criminal procedure. This is because the assumptions it makes about the
relative frequency and costs of false convictions and false acquittals are
implausible in light of the empirical evidence.279
We provide another rationale for the strong pro-defendant bias in criminal
procedure and argue that this provides a more complete justification for the
extent of the pro-defendant bias in criminal procedure. We argue that the strong
pro-defendant bias is justified as a means to constrain the costs associated with
self-interested behavior by prosecutors and government agents. Absent some
constraint on their behavior, prosecutors and other agents might be tempted to
use the criminal process to benefit themselves. This prospect is likely to induce
various groups in society to lobby prosecutors and other government officials for
selective enforcement. The direct lobbying, counter-lobbying, and associated
costs might be quite large. When these costs are combined with the deleterious
effects on deterrence such lobbying could have, we find that the costs of
(finding results consistent with the notion that the
common law leads to greater economic growth relative to civil law systems in the period 1960 – 1992).
278 See Marshall & Duff, supra note 3; Estrick, supra note 3; Hart, supra note 3. See also Coffee, supra
note 3; Kadish, supra note 3; PACKER, supra note 3.
279 See supra Part III.C.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 61
unfettered prosecutorial behavior could plausibly be large enough to justify pro-
defendant procedural protections.
After discussing why these constraints, as opposed to others one could
imagine, are likely to be useful, we assess whether the Reasonable Doubt
Standard, Double Jeopardy and the Right to Jury Trial reflect these concerns. We
find there is much in broad contours in these areas that is consistent with our
analysis. To test this further we also examine some of the more minute details of
case law related to Double Jeopardy, Right to a Jury Trial, Ex Post Facto
Prohibitions, Void-for-Vagueness Doctrine, Entrapment, and Penalty
Restrictions. We find that our approach helps explain several details that are
difficult to explain under traditional theories.
We then consider whether there is empirical evidence to support our
approach and find indirect support in corruption data from several countries. 280
This suggests that approaching the area of Criminal Procedure from the
perspective of constraining self-interested prosecutors is likely to provide
important insights into our current scope of Criminal Procedure as well as
insights about whether certain doctrines should be extended or not. Indeed, the
analysis developed here could be applied to a myriad of current topics, including
the extension of criminal procedural protections to civil suits brought by
government agencies.281
In the end, the goal of our paper is to provide an economic theory that
might provide some grounds for justifying and assessing American Criminal
Procedure. We argue that attempting to constrain the costs associated with self-
interested prosecutors or government agents using the criminal process to
280 See supra Part VII.D.
281 The government may often bring civil suits or administrative proceedings against parties and
similar fears of rent-seeking may arise in those contexts. See Kenneth Mann, Punitive Civil Sanctions: The
Middleground Between Criminal and Civil Law, 101 YALE L.J. 1795, 1797, 1862-63 (1992) (discussing the role of
punitive civil sanctions). We do not, however, witness the same degree of pro-defendant protections in
these areas. See id. at 1869-70. Our analysis does not examine whether these fields should have stronger pro-
defendant procedures as that is a detailed and lengthy topic worthy of at least another paper. See generally
id.; Mary M. Cheh, Constitutional limits on Using Civil Remedies To Achieve Criminal Law Objectives:
Understanding and Transcending the Criminal-Civil Law Distinction, 42 HASTINGS L.J. 1325 (1997); Carol S.
Steiker, Punishment and Procedure: Punishment Theory and the Criminal-Civil Procedural Divide, 67 GEO.WASH.
L. REV. 1290, 1292 (1997). We would note at this point that there are some differences between government
civil suits and criminal enforcement. First, the sanctions and stigma are often different in both contexts and
this may play a role in determining how pro-defendant the procedures need to be in these different contexts.
See Mann, supra, at 1809. Second, we may believe that the potential defendants in government civil cases
(e.g., corporations) are able to lobby more effectively than the average defendant in the criminal context.
This is because relatively few criminal defendants are wealthy enough to even litigate effectively, suggesting
that the scope of their lobbying abilities are somewhat limited. See Stuntz, supra note 14, at 28 – 29.
However, these are only preliminary thoughts – our current paper does not engage in the debate over
whether government civil suits should be subject to heightened procedures. This is a matter left for future
debate and analysis.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 62
benefit themselves provides such a justification. Our approach also seems
consistent with case law and with the empirical evidence we currently possess.
In light of this, it would seem important to bring explicit consideration of our
approach into the calculus when assessing current Criminal Procedure and
discussing how, if at all, it should be changed.
Toward an Economic Theory of Pro-Defendant Criminal Procedure 63
Appendix
Below we report the results of expanded “corruption” regressions. The
new variables below are EDU1 = percentage of population (25 and older in 1991)
that has something less than a primary education only. This includes those who
have no education at all or have completed primary school but have gone no
further. (Source: Statistical Abstract of the World, 3d. ed., Annmarie Muth, ed.,
Gale Research (1997); RELC = percentage of the population who are Catholic,
RELM = percentage of the population who are Muslim or Islamic (Source:
Statistical Abstract, and where necessary supplemented from
http://www.adherents.com/). ETM = dummy equal to one if economy is classified as
mixed socialist-capitalist, ETS = dummy equal to one if economy is classified as
socialist (Source: http://www.cia.gov/cia/publications/factbook/indexgeo.html).
Table A.1
CPI Coef. t Coef. t
RATIO .045 1.611 .047 1.979
CRUPUN3 -1.866 -2.790 -2.498 -4.352
CRUPUN2 -1.736 -2.563 -2.128 -3.629
COMMLAW 1.184 2.047 .921 1.772
EDU1 -.028 -2.103
RELC -.009 -1.163 -.014 -2.028
RELM -.006 -.551 -.012 -1.380
ETM -.843 -1.394 -.430 -.815
ETS -1.668 -2.194 -1.615 -2.423
CONS 4.986 6.832 5.813 7.355
Number 61 75
R-squared .535 .522